2015-09-01 06:37:32 Sleaker: we do it for free 2015-09-01 06:37:39 we dont sell support 2015-09-01 06:39:51 iirc, APK_REPOS is so you can add a non-standard repository whne you create images 2015-09-01 06:40:01 the packages there will just be added to the repo on boot media 2015-09-01 06:40:22 nothing is installed at boot time unless you ship an *.apkovl.tar.gz file 2015-09-01 07:58:37 Sleaker: how are you getting on with your alpine linux setup? 2015-09-01 08:52:18 why isn't there any sponsors for AL ? 2015-09-01 08:58:33 I don't think selling support means current state of free support, it comprises more of "24x7 expert paid services" 2015-09-01 08:59:45 vkris, AL is mostly sponsored by offering people to work on it 2015-09-01 08:59:50 may be start "integration services" for a fee and maybe divert $200 thats running short in AL's budget 2015-09-01 09:00:23 most of the main developers do not have capacity to offer 24x7 support 2015-09-01 09:00:50 that is why it would be paid 2015-09-01 09:01:20 when that is possible AdHoc developers and be on standby 2015-09-01 09:01:29 can be* 2015-09-01 09:01:58 if fee becomes incentive more developers can join 2015-09-01 09:08:32 ]/win 32 2015-09-01 12:10:40 looks nice, https://allseenalliance.org/ 2015-09-01 13:22:50 Will alpine use systemd? 2015-09-01 13:22:58 :| 2015-09-01 13:23:03 http://forum.alpinelinux.org/forum/general-discussion/alpine-plans-systemd ! 2015-09-01 13:30:15 frankS2, isn't the answer there? 2015-09-01 13:30:29 Yep I found it, just pasted it for anyone else who might wonder :) 2015-09-01 13:30:43 Yay to no-systemd 2015-09-01 13:33:13 My current Arch installation is borked because of systemd 2015-09-01 13:33:45 Network doesn't always come up properly and when I shut down, it waits *infinitely* for br0 to stop 2015-09-01 13:34:20 CandyAngel: since you're talking about it just now: 2015-09-01 13:34:35 "warning: systemd: ignoring package upgrade (219-6 => 224-1)" :P 2015-09-01 13:34:44 And I'm pretty sure it has broken the sound too 2015-09-01 13:35:34 Basically, whenever something goes wrong, it's mostly systemd or pulseaudio. I'm not even kidding >.< 2015-09-01 13:35:52 I didn't even know systemd was Lennarts project, but I wasn't that surprised when I found out :P 2015-09-01 13:35:56 np 2015-09-01 13:36:07 bah...wrong window 2015-09-01 13:36:14 I also get to add annoying "$ pacman -Syu --assume-installed libgudev" extras all the time because of systemd everything-in-a-box (except it doesn't work) mentality 2015-09-01 13:36:22 :P 2015-09-01 13:36:45 Once my new VM setup is working, I'll probably try out Gentoo for my main Linux VM 2015-09-01 13:37:00 I actually watched a talk at some BSD conforence about porting Apple's launchd to FreeBSD 2015-09-01 13:37:20 Not familiar with launchd. Is it any good? 2015-09-01 13:37:22 Apple seems to have gotten a lot of things right that systemd craps out on >.< 2015-09-01 13:37:35 It runs OS X since like 10.6 2015-09-01 13:37:51 I don't think there's a Linux port though. It depends on Mach IPC 2015-09-01 13:37:57 Ah 2015-09-01 13:38:07 Well, more choice is good so I hope they can port it :) 2015-09-01 13:38:12 The FreeBSD guys just added Mach IPC to their kernel as a module 2015-09-01 13:38:22 Oh, nice 2015-09-01 13:38:51 I was actually considering having a look at launchd porting myself, but I don't have the time >.< 2015-09-01 13:39:49 The talk I was referencing: http://nextbsd.org/jordan-hubbard-visits-bafug/ 2015-09-01 13:41:09 The key notion I love about launchd is that lanunchd itself (PID1) should be as minimalistic as possible 2015-09-01 13:41:43 Something I don't dare to say about systemd 2015-09-01 13:44:30 Yeah, I don't know a lot about that sort of thing but I always imagined the first process to be *super* minimal and pass off as much to something else as possible 2015-09-01 13:46:21 But yeah, I got very jealous of how smart and clever other people are. I wish I was that smart :P 2015-09-01 13:46:56 is is minimal? 2015-09-01 13:47:15 fsr it has an xml parser somewhere 2015-09-01 13:47:23 It doesn't 2015-09-01 13:47:29 launchctl does the parsing 2015-09-01 13:47:34 not sure whether that's in the pid1 binary though 2015-09-01 13:47:38 okay good 2015-09-01 13:48:27 launchd basically calls "launchctl --bootstrap" when it's ready and that collects the service definitions, parses them and tells launchd about it 2015-09-01 13:48:47 My biggest contribution (aside from helping people where I can) is a Perl wrapper for a module that 2 people use :| 2015-09-01 13:49:12 And it's not even on CPAN yet 2015-09-01 13:49:18 As long as it's useful to you ;P 2015-09-01 13:50:22 I think I really need to reboot D: 2015-09-01 13:50:33 Well, I means I can add VR support to my games :P 2015-09-01 13:50:37 it means* 2015-09-01 13:51:15 I love how I can batch-update all my lxc containers :D 2015-09-01 13:51:50 BY restarting? :P 2015-09-01 13:52:02 no, BEFORE restarting ;) 2015-09-01 13:52:47 They also share the package cache, so no duplicate package downloads and duplicate packages taking up space :) 2015-09-01 13:53:10 :) 2015-09-01 13:53:18 Alpine in LXC containers? 2015-09-01 13:53:28 Mostly, yeah 2015-09-01 13:53:56 A couple Alpine and one Arch 2015-09-01 13:54:32 Coolies 2015-09-01 13:54:47 The batch update thing doesn't work with Arch as guest though 2015-09-01 13:54:52 pacman doesn't like lxc-attach 2015-09-01 13:56:16 https://github.com/Orochimarufan/lxc-tools/blob/master/bin/lxc-update#L33 2015-09-01 13:57:49 That's a shame 2015-09-01 13:57:57 It's a known issue 2015-09-01 13:58:47 So is my touchpad freaking out and losing connection on my laptop. Pacman/LXC might get fixed :P 2015-09-01 13:58:59 hmm actually, they might have fixed it :O let me see 2015-09-01 13:59:27 Once it's done compiling wine, that is :P 2015-09-01 14:00:01 ...what 2015-09-01 14:00:01 Got me there, algitbot :P 2015-09-01 14:00:13 I just found out I can click-drag in Putty and it changes the weechat window 2015-09-01 14:00:36 I haven't used weechat in a while^^ 2015-09-01 14:01:03 magic 2015-09-01 14:01:35 Also, I've moved away from windows as main OS so I don't use putty that often anymore either 2015-09-01 14:04:41 I have to use Windows at work :( 2015-09-01 14:05:06 I don't work (yet) 2015-09-01 14:05:42 Nope, pacman is still broken: 2015-09-01 14:05:44 error: GPGME error: Inappropriate ioctl for device 2015-09-01 14:05:44 error: lib32-jack2: missing required signature 2015-09-01 14:05:44 :: File /var/cache/pacman/pkg/lib32-jack2-1.9.10-2-x86_64.pkg.tar.xz is corrupted (invalid or corrupted package (PGP signature)) 2015-09-01 14:08:50 Orochimarufan, try with | cat 2015-09-01 14:09:16 I see what you're suggesting 2015-09-01 14:09:52 It actually works... 2015-09-01 14:11:03 You want to do# lxc-attach -n -- sh -c 'pacman -S | cat' 2015-09-01 14:11:19 (Note the sh -c part) 2015-09-01 14:11:33 ncopa: http://forum.alpinelinux.org/forum/pax-grsecurity/grsec-patches-only-sponsors 2015-09-01 14:12:43 Is there a reason why there is no official torrent mirror of Alpine? 2015-09-01 14:12:58 frankS2: yes. it affects alpine 2015-09-01 14:13:17 Orochimarufan, heh your google-fu isn't as sharp as it should :P 2015-09-01 14:13:19 we need to ask spender what options we have 2015-09-01 14:13:52 jvoisin: only reason is that we havent found a convenient way to autocreate the torrent yet 2015-09-01 14:14:05 TheLemonMan: it's not on the arch wiki, and back when I did a thourough search it didn't seem to be there (that's some months back though) 2015-09-01 14:14:43 Thanks, anyway 2015-09-01 14:15:35 ncopa: I know that Tails is usink mktorrent ( https://github.com/mukaibot/mktorrent ) 2015-09-01 14:15:49 Orochimarufan, feel free to add it to the wiki 2015-09-01 14:15:51 And we're happy with it 2015-09-01 14:16:33 it'd be nice to know if there's a way to avoid this hack, from what I remember it's because gpgme sends an ioctl to the tty 2015-09-01 14:16:53 I thought something like that was the issue 2015-09-01 14:17:19 Apparently, lxc-attach does some voodoo with ttys and pts 2015-09-01 14:17:44 I don't really know how the whole lxc-attach business work behind the curtains, but it might be easy to fix *wink* 2015-09-01 14:18:15 There's so many things that would probably be easy to fix if someone found the time to do so^^ 2015-09-01 14:22:57 found alternative for nodejs, https://github.com/creationix/dukluv, nice from AL point 2015-09-01 14:23:14 I've started writing a wiki page regarding using Alpine as KVM host ^_^ 2015-09-01 14:23:27 I hope it isn't too specific to how I want it to be set up when it's done 2015-09-01 14:24:14 I use Arch as the host 2015-09-01 14:24:36 With both QEMU/KVM/libvirt and LXC running on top 2015-09-01 14:25:09 Mine is(/will be) Alpine with QEMU/KVM and VGA passthrough for 2 of the VMs 2015-09-01 14:25:29 The host is fairly minimal though, which is good 2015-09-01 14:25:42 I do something similar with LXC 2015-09-01 14:25:57 I'm running my desktop inside the sole Arch Container 2015-09-01 14:26:16 Coolies 2015-09-01 14:26:28 It works pretty much out of the box once you have UDev forwarding figured out 2015-09-01 14:26:45 (UDev doesn't work inside containers) 2015-09-01 14:26:54 It might be that I could use Arch (which would make sense, as I've used it) 2015-09-01 14:27:05 But I liked the sound of Alpine :P 2015-09-01 14:27:41 Though a bit of a problem with vfio I think.. it's slightly too old? 2015-09-01 14:27:58 I tried using vfio-pci but it said 'ids' was an unknown parameter 2015-09-01 14:28:08 So I think it needs kernel 4+ 2015-09-01 14:28:10 IMHO, Alpine seems better suited to "serving" workloads than actual desktop use 2015-09-01 14:28:24 That's why it's my KVM host :) 2015-09-01 14:28:32 I use it to run all my network services 2015-09-01 14:28:50 (Inside containers ofc) 2015-09-01 14:28:54 Did want to use it for a really small VM which just has a couple of programs I *always* use, but neither of them are packaged :( 2015-09-01 14:29:01 ScrumpyJack: doing well 2015-09-01 14:29:16 Alpine packaging is actually really straight forward 2015-09-01 14:29:43 My plan was a small VM with Leo, Gnucash and a browser (any would probably be fine) 2015-09-01 14:29:54 Then Windows VM with passthrough to my GTX 570 (for games) 2015-09-01 14:30:11 For the most part, you can usually take a PKGBUILD from AUR/ABS and do some manual plumbing and there you go 2015-09-01 14:30:15 And a Linux VM with passthrough to a GT 610 for all the other stuff I do in Linux :P 2015-09-01 14:30:48 Personally, I'd say use KVM only for the Windows one and use containers for the linux ones 2015-09-01 14:31:11 But that comes with a couple of corner-cases that you'd have to be prepared to deal with :( 2015-09-01 14:31:27 (It's what I do so I somewhat know what I'm talking about) 2015-09-01 14:31:46 Well, I'm always up for suggestions and improvements to my plan :P 2015-09-01 14:32:42 Well, if you decide to give lxc a shot (is it packaged in Alpine?), feel free to bug me about it (Provided I'm around ofc) ;) 2015-09-01 14:32:53 Thankies, I really appreciate the offer 2015-09-01 14:33:25 And it seems to be packaged, so there's that 2015-09-01 14:34:02 The main problem with the programs I want in the small VM is that it's Leo (required PyQt) and GnuCash (probably requires a bunch of GNOME stuff) 2015-09-01 14:34:32 I run a full KDE desktop (on Arch), so technically that shouldn't be a problem 2015-09-01 14:34:43 You might have to do some packaging work for Alpine is all 2015-09-01 14:35:24 CandyAngel: virtualbox? 2015-09-01 14:35:43 Sleaker: QEMU/KVM 2015-09-01 14:35:52 do they support hardware passthrough? 2015-09-01 14:35:56 Yeah 2015-09-01 14:36:35 I guess vbox uses qemu as a backend anyway 2015-09-01 14:36:36 My friend bought the same hardware I have to test it out and he has Windows VM with passthrough working (~95% of native performance in games) 2015-09-01 14:36:43 yah 2015-09-01 14:36:59 I don't think vbox uses qemu... 2015-09-01 14:37:06 I'm pretty sure it doesn't 2015-09-01 14:37:15 https://www.virtualbox.org/wiki/Developer_FAQ 2015-09-01 14:37:46 I guess they just took code from QEMU is all 2015-09-01 14:37:52 Yeah 2015-09-01 14:38:38 It might have components based on QEMU, but it doesn't "use" it 2015-09-01 14:38:58 yah I thought they had actually allowed qemu as a backend. guess not. 2015-09-01 14:39:12 My main problem with QEMU is the really poor performance when the VM display is >1024x68 2015-09-01 14:39:13 Nah, they have their own stuff 2015-09-01 14:39:32 Even with QXL, it is really bad (hence buying the GT 610 for the main Linux VM) 2015-09-01 14:39:49 I guess 2015-09-01 14:39:52 Le cheap option, so hopefully it is enough 2015-09-01 14:40:00 (it was like.. £30) 2015-09-01 14:40:07 Shouldn't there be a VirtIO graphics channel these days? 2015-09-01 14:40:49 also, you realize you need some kind of muxer/switcher to plug both GPU outputs into the same screen, right? 2015-09-01 14:41:20 Have you thought about using xen instead of KVM? 2015-09-01 14:41:27 That's one of the things I dislike about passthrough at this point 2015-09-01 14:41:48 ScrumpyJack: What benefits would it give him? 2015-09-01 14:41:48 there is a nice alpine-xen iso to get you started with alpine as your Dom0 2015-09-01 14:42:00 Orochimarufan: ^^ 2015-09-01 14:42:10 I guess that's an argument 2015-09-01 14:42:55 are the display screens remote? 2015-09-01 14:43:04 vnc/spice etc 2015-09-01 14:43:25 No, he wants to use pci(e) passthrough for a dGPU 2015-09-01 14:45:51 http://wiki.xen.org/wiki/Xen_PCI_Passthrough 2015-09-01 14:45:54 OKay 2015-09-01 14:46:13 I can't use Xen because my GPU can't be passed to it 2015-09-01 14:46:20 (not multi-OS and can't be "hacked to be) 2015-09-01 14:46:27 I was actually looking at Xen first 2015-09-01 14:46:34 How I have it set up currently is: 2015-09-01 14:46:49 CPU-GPU: HDMI to monitor 2015-09-01 14:46:49 I think your only options for passthrough are KVM and VBox 2015-09-01 14:46:50 passthrough should work the same between xen and qemu :O 2015-09-01 14:46:56 GTX 570: DP to monitor 2015-09-01 14:47:07 GT 610: DVI to monitor 2015-09-01 14:47:36 The only connection I'm not using on the monitor is VGA :P 2015-09-01 14:48:05 heh 2015-09-01 14:48:22 Mine only has HDMI and VGA >.< 2015-09-01 14:48:51 I got a cheap switcher to switch the HDMI input from 3 PCs and my PS3 2015-09-01 14:49:04 It's everything but a good setup 2015-09-01 14:50:05 Regarding QXL and SPICE, I *think* I've used it when I'm messing about in Arch and it still performed poorly 2015-09-01 14:50:17 so you want to pass pci device xx:xx.xx to one guest and pci device yy:yy.yy to another, and use your monitor to switch between inputs? 2015-09-01 14:50:19 BUt I'm not sure it was even working correctly, because my Arch install is messed up 2015-09-01 14:50:42 For example, I have to start virt-manager as root because I can't get it to talk to qemu: or whatever otherwise 2015-09-01 14:50:47 ScrumpyJack: Exactly 2015-09-01 14:51:06 QXL is SO much better than the other stuff where it works though 2015-09-01 14:51:16 I'm sure it is 2015-09-01 14:51:21 Unfortunately, it only works up to Win7 2015-09-01 14:51:42 I can't even my Alpine VM to show X unless I'm using QXL :| 2015-09-01 14:51:53 And it works fine at 800x600 :P 2015-09-01 14:52:22 Noticeably laggy at 1280x720 2015-09-01 14:52:26 Don't know about Linux guests, I don't use KVM for that 2015-09-01 14:52:38 jvoisin: thanks for sharing 2015-09-01 14:52:52 sounds good to me 2015-09-01 14:52:52 Waiting for like.. 30 seconds for full window switches at 2560x1440 (native monitor) 2015-09-01 14:53:14 It's great 2015-09-01 14:53:16 jvoisin: i suppose we need someone to start seed it too 2015-09-01 14:53:23 (Well, except that it's ruby ofc) 2015-09-01 14:53:27 :) 2015-09-01 14:53:29 ofc :) 2015-09-01 14:53:56 You can install a simple torrent client on your server I guess 2015-09-01 14:53:57 it would be interesting to see it working. keep a diary of your adventures CandyAngel :) 2015-09-01 14:54:24 transmission is a nice headless torrent client IMO 2015-09-01 14:54:35 ScrumpyJack: I am doing. I'm writing a guide up on the wiki 2015-09-01 14:54:46 hmmm. wonder if I could create an overlayfs for the USB boot option and have all the OS packages, etc, install over the top onto that, to simulate a live env. 2015-09-01 14:54:52 err to simulate a sys install. 2015-09-01 14:55:15 ScrumpyJack: http://wiki.alpinelinux.org/wiki/User:CandyAngel/KVM_Host_Tutorial 2015-09-01 14:55:36 Just a command dump right now, I'll actually make it a proper guide once it is working :P 2015-09-01 14:55:43 ncopa, is overlay supported in the alpine kernel? 2015-09-01 14:56:11 CandyAngel: y u use ext3 :O 2015-09-01 14:57:31 Also, why run X on your host? 2015-09-01 14:57:34 Because it works? 2015-09-01 14:57:48 For the non-VGA passthrough VMs I will have 2015-09-01 14:58:13 But then you'd have to hook up 3 thing to the screens 2015-09-01 14:58:27 I already do 2015-09-01 14:58:32 I see 2015-09-01 14:58:42 HDMI is host, DP will be Windows, DVI will be Linux 2015-09-01 14:59:06 you could also use spice to display the non-pass vms from your linux vm :P 2015-09-01 14:59:22 But then I can't use the non-pass VMs unless I have my Linux VM up 2015-09-01 14:59:27 I want them to be separate :P 2015-09-01 14:59:28 Anyway, do you have any reason to use ext3 over ext4 at this point? 2015-09-01 14:59:55 Last time I used ext4, it messed up my files because of stupid defaults 2015-09-01 15:00:17 When was that? 2015-09-01 15:00:33 Ages ago 2015-09-01 15:00:38 Thought so 2015-09-01 15:00:41 Like.. a few months after it came out? 2015-09-01 15:00:56 There's a proposal to remove the ext3 driver from Linux 4.3 2015-09-01 15:01:02 And have the ext4 driver handle it 2015-09-01 15:01:05 However, the number of times I've reinstalled my OS since I switched back to ext3 is.. 0 2015-09-01 15:01:14 (ignoring my new VM setuping) 2015-09-01 15:01:42 What I'm saying is, it's considered stable now and you should consider creating any new filesystems as ext4 2015-09-01 15:01:59 (It's actually been considered stable for a long time now) 2015-09-01 15:02:09 Part of the reason I want it setup like this is so it is fairly easy to swap out without spending ages reinstalling 2015-09-01 15:02:10 yah I've never had problems with ext4 2015-09-01 15:02:25 For example, even if I reinstall Alpine on ext4 as host.. that'll be like.. 15 minutes? 2015-09-01 15:03:01 Personally, I mostly use btrfs these days, though I can see how people are still weary of it 2015-09-01 15:03:18 Yeah, I'm not saying it is unstable or crap, I just haven't had chance to give it another chance is all 2015-09-01 15:03:41 Well, now's your chance! 2015-09-01 15:03:45 And I don't really want to give it that chance where my host is :P 2015-09-01 15:03:49 My Linux VM can be ext4 2015-09-01 15:04:21 :P 2015-09-01 15:04:27 Well, these commands are for the vm, right? :P 2015-09-01 15:05:02 It's for going from blank bare metal to having Windows/Linux VMs with passthrough 2015-09-01 15:05:29 yup, and you use mkfs.ext3 as part of it 2015-09-01 15:05:42 Mhm, for /boot and / for the KVM host (Alpine) 2015-09-01 15:06:51 I was under the impression a minimally set-up alpine host systemd was prerequisite 2015-09-01 15:07:04 huh? 2015-09-01 15:07:13 why would you taint a perfectly good distro with systemd 2015-09-01 15:07:15 wow epic type 2015-09-01 15:07:21 *wow epic typo 2015-09-01 15:07:27 lol 2015-09-01 15:07:34 s/systemd/system/ 2015-09-01 15:07:53 it's not even in the repo :P 2015-09-01 15:08:01 Good riddance 2015-09-01 15:08:13 I'm not sure what you mean? 2015-09-01 15:08:32 CandyAngel: I'll just let you finish that article and THEN start commenting ;) 2015-09-01 15:08:50 Sleaker: I already posted this earlier, but: 2015-09-01 15:08:51 lol 2015-09-01 15:08:59 warning: systemd: ignoring package upgrade (219-6 => 224-1) 2015-09-01 15:09:08 ^ only sane thing to do on tainted distros 2015-09-01 15:09:26 It gets worse every release 2015-09-01 15:09:36 The guide is to take you through the whole process from a computer with 2 blank HDDs in it to how I have it set up 2015-09-01 15:09:59 So it combines things from the GPT LVM and RAID1 articles too 2015-09-01 15:10:26 I thought 204 was the last 'sane' release 2015-09-01 15:10:31 or was it 209? 2015-09-01 15:10:37 who knows. whichever one is on 14.04 2015-09-01 15:10:46 Ubuntu( 2015-09-01 15:10:53 that'd be 204 2015-09-01 15:11:14 but 205-219 were usable at least 2015-09-01 15:11:29 with 220 they started going completely nuts 2015-09-01 15:11:33 Hm 2015-09-01 15:11:43 I thought everything after 204 replaced syslog. 2015-09-01 15:11:52 it doesn't replace syslog 2015-09-01 15:11:53 If I remember correctly, 213 had a bug where it just flat out didn't even try to bring your network up :| 2015-09-01 15:11:59 Or something like that 2015-09-01 15:12:00 lol 2015-09-01 15:12:08 it has it's own logging system that can work alongside syslog 2015-09-01 15:12:19 (And also forward all its messages TO syslog) 2015-09-01 15:12:28 sounds like cruft 2015-09-01 15:12:36 only need 1 logging system 2015-09-01 15:12:51 the Journal isn't that bad really, except why does it have to come bundled with SD 2015-09-01 15:12:52 Hm 2015-09-01 15:13:08 On the wiki, is it worth having little warnings for "expected, but harmless" error messages 2015-09-01 15:13:38 Like.. I have to put mdadm-raid at the boot runlevel otherwise /boot doesn't get mounted, but it makes Alpine complain while booting 2015-09-01 15:13:39 There are quite a few good ideas in the sd project, the problem is them intermingling everything 2015-09-01 15:14:19 The problem is them thinking anyone who disagrees with them is wrong.. 2015-09-01 15:14:32 That too 2015-09-01 15:14:45 Anyone thinking that is a retard 2015-09-01 15:18:13 Except me of course. I'm always right B) 2015-09-01 15:18:22 (I jest, hopefully obviously :P) 2015-09-01 15:19:45 I think I am going to try again and get vfio-pci or pci-stub working tonight 2015-09-01 15:19:54 Prefer that to a script which does a bunch of unbinding and such 2015-09-01 15:42:50 CandyAngel: This could be interesting: https://www.phoronix.com/scan.php?page=news_item&px=AMD-Hardware-GPU-Virt 2015-09-01 15:43:15 anyway, need to reboot brb 2015-09-01 15:48:19 hmmm I'm confused as to why people don't consider using alpine as a desktop, even though it's not necessarily marketed as such 2015-09-01 15:56:19 bonsoir 2015-09-01 15:57:39 hi 2015-09-01 15:58:31 Heyas TsT 2015-09-01 15:58:44 Sleaker: Because it is missing a lot of stuff so you can't "just install" 2015-09-01 15:58:54 That's the reason for me, anyway 2015-09-01 15:58:57 hmmm like? 2015-09-01 15:59:37 Let's see what software I use that isn't there.. 2015-09-01 15:59:40 Leo 2015-09-01 15:59:42 GnuCash 2015-09-01 15:59:44 Blender 2015-09-01 16:00:09 git-annex 2015-09-01 16:00:21 perlbrew 2015-09-01 16:01:21 hmm ok. 2015-09-01 16:01:59 Steam (eugh) 2015-09-01 16:02:18 Teamspeak :P 2015-09-01 16:02:39 steam has no source version, would it even run on musl? 2015-09-01 16:02:57 Those are the things that come to mind anyway 2015-09-01 16:03:32 I think you've have to have a very particular usage for Alpine to "just work" as a desktop 2015-09-01 16:03:40 In terms of package availability I mean 2015-09-01 16:03:43 ah yah 2015-09-01 16:03:45 that's valid 2015-09-01 16:04:29 And from the looks of it, that is "very general" things 2015-09-01 16:04:53 And anyone who is enough of Linux power user to want Alpine is going to go out that general zone pretty quick 2015-09-01 16:05:03 right 2015-09-01 16:05:20 From my quick look, it would suit fine for installing for someone to browse the web 2015-09-01 16:05:28 And stuff like that 2015-09-01 16:05:42 well yah, just anything with xfce handles that 2015-09-01 16:05:46 Yeah 2015-09-01 16:30:13 ncopa - what's up with chrony locking up on start for like 3-4 seconds? 2015-09-01 16:33:32 ncopa, is arm missing some packages that are present on x86? 2015-09-01 16:33:46 i'm trying scaleway and: 2015-09-01 16:34:01 biceps:~# apk add emacs 2015-09-01 16:34:02 ERROR: unsatisfiable constraints: 2015-09-01 16:34:02 emacs (missing): 2015-09-01 16:34:10 emacs isn't in normal 2015-09-01 16:34:18 just looked for it myself 2015-09-01 16:34:22 i have edge/main and edge/testing 2015-09-01 16:34:29 ah k 2015-09-01 16:35:09 not in testing for me either 2015-09-01 16:35:48 dalias: I'm not seeing emacs in the repo at all 2015-09-01 16:36:01 i have it on x86_64 2015-09-01 16:36:11 maybe it's only for that arch, I'm on x86 2015-09-01 16:36:29 http://git.alpinelinux.org/cgit/aports/tree/testing/emacs 2015-09-01 17:05:43 alright, I'm back 2015-09-01 17:45:50 anyone still wants antivirus, http://www.alphr.com/security/1001436/can-we-trust-antivirus-companies-anymore 2015-09-01 17:46:59 ncopa, how do i install the kernel on alpine/arm? do the normal kernel packages work? 2015-09-01 17:47:57 installing apline on scaleway gave me a 3.2 kernel which is perhaps provided by them rather than by alpine..? 2015-09-01 18:54:46 Gah 2015-09-01 18:54:54 This is why I don't trust Alpine 2015-09-01 18:55:04 I just cloned the partitions from one drive to another 2015-09-01 18:55:15 Something I've done a load of times before (in Alpine no less) 2015-09-01 18:55:25 This time, I only get /dev/sde1 2015-09-01 18:55:33 Instead of /dev/sde1 and /dev/sde2 :| 2015-09-01 18:57:41 This is the second time I've executed the same series of commands and had Alpine do something weird 2015-09-01 19:03:15 which commands? 2015-09-01 19:03:33 did you cloned the partition tables as well? 2015-09-01 19:05:41 Yeah 2015-09-01 19:05:51 I use gdisk to create 2 partitions on /dev/sdb 2015-09-01 19:05:54 Then I clone it using 2015-09-01 19:06:02 sgdisk /dev/sdb -R /dev/sde 2015-09-01 19:06:07 sgdisk -G /dev/sde 2015-09-01 19:06:14 Normally, this works fine 2015-09-01 19:09:24 Huh 2015-09-01 19:09:40 It did it again, after the second bit (it has both first) 2015-09-01 19:09:55 So the second one is vanishing when I randomise the GUID's 2015-09-01 19:10:02 But it didn't last time >.< 2015-09-01 19:11:01 If I rewrite it in gdisk, it works 2015-09-01 19:11:04 :> 2015-09-01 19:11:06 :| 2015-09-01 19:12:44 ACTION is now confused 2015-09-01 19:17:56 Annnd I did it again and it worked fine that time 2015-09-01 19:17:58 -.- 2015-09-01 19:31:30 I think basing a whole article around an virus that affected ~100 users and saying that AV is colluding is rediculous 2015-09-01 19:31:59 AV job is to do exactly what the article described in the middle, to prevent widespread consumer virus from infecting people. 2015-09-01 19:32:56 it doesn't make economic sense to target a super advanced piece of malware that has such a super low target to begin with when you want to sell your product to a mass of people. If you happen to figure out how it works and add in preventative measures than that's great. 2015-09-01 19:33:15 I think it's a bit belittling to the companies to try and state that they are being malicious or colludin though. 2015-09-01 19:33:26 ^ vkris 2015-09-01 21:01:09 Hm 2015-09-01 21:01:37 is /etc/modules only for modules that are needed during boot or is it for things that can be started later? 2015-09-01 21:06:51 Nevermind 2015-09-01 21:07:04 I guess it goes in /etc/modules-load.d :P 2015-09-01 21:12:10 :) 2015-09-01 21:47:48 hum... i think there are issues with the golang 1.5 package :( 2015-09-01 21:49:41 i've got a project that includes go-sqlite, if i compile it with -race it directly bails out, if i compile it normal with it, it segfaults, if i remove go-sqlite it just works 2015-09-01 22:36:27 What is the modules entry in APPEND for that is different to just adding things to /etc/modules? 2015-09-01 23:18:40 Argh, I can't get pci-stub to work -.- 2015-09-01 23:39:22 vfio problems? 2015-09-01 23:40:40 Kind of 2015-09-01 23:41:21 I can't get pci-stub to take the devices instead of nouveau/snd_hda_intel 2015-09-01 23:41:31 I've added it to /etc/modules 2015-09-01 23:41:42 mkinitfs'd 2015-09-01 23:41:59 And added pci-stub.ids=aaaa:bbbb,aaaa:cccc 2015-09-01 23:42:06 to my APPEND line 2015-09-01 23:43:37 But nouveau and snd_hda_intel still get them 2015-09-02 00:10:35 I don't know what else to try :/ 2015-09-02 00:11:22 The module is getting loaded 2015-09-02 00:11:40 Doesn't show any parameters in /sys/module/pci_stub.. 2015-09-02 07:11:03 dalias: i think you shoudl be able to apk add linux-grsec? 2015-09-02 07:11:12 or apk add linux-vanilla 2015-09-02 07:11:24 but idont have that much experience with arm 2015-09-02 07:13:25 ncopa: you should power up that RPi you have ;) 2015-09-02 07:14:24 :) 2015-09-02 07:14:38 i have it right here 2015-09-02 07:15:20 anyone used seafile with alpine linux? 2015-09-02 07:15:44 i tried to use it but never made it work 2015-09-02 07:15:53 never figured out how it was supposed to be package 2015-09-02 07:16:09 it seem like thye hhave made it so you can git clone && run 2015-09-02 07:16:18 but not repack it 2015-09-02 07:16:49 i thimk maybe [xming] has made it run? 2015-09-02 07:18:17 there seems to be a package in the repos now 2015-09-02 07:19:10 and your down as maintainer :) 2015-09-02 07:19:38 http://wiki.alpinelinux.org/wiki/Seafile:_setting_up_your_own_private_cloud 2015-09-02 07:34:49 Do options on the kernel command line only get passed to built-in modules or do they get passed to loaded modules too? 2015-09-02 08:38:59 what's the minimum RAM reqs for alpine linux i386? anyone tested? 2015-09-02 09:09:07 ScrumpyJack: I think I went down as far as 32MB, but have not tried less 2015-09-02 09:29:26 royger: alpine-mini-x86 needs a 64bit CPU, is that correct? 2015-09-02 10:10:38 yeah, 32m is the minimum 2015-09-02 10:59:26 on a sys install, can i set up apkcache if i just want to cache packages for apk add ? 2015-09-02 11:45:36 i'm a bit confused with the alpine cache 2015-09-02 11:45:55 when i apk add moo, does it cache moo as well as install it? 2015-09-02 11:52:26 what i'm trying to do is create an lxc alpine container on an alpine host without downloading the packages from the network 2015-09-02 11:59:40 ScrumpyJack: Isn't that how Orochimarufan has it set up? 2015-09-02 12:34:57 Orochimarufan: you about? 2015-09-02 12:35:51 there are some tweaks to do this but i'm not quite sure how to do them all 2015-09-02 12:36:14 mainly add local dir to /etc/apk/repository 2015-09-02 12:36:57 but then an apkindex file needs to be created 2015-09-02 12:38:05 the apk inside the apline lxc template is where i'm trying to pull from cache 2015-09-02 12:40:32 i tried apk index but got lots of missing deps 2015-09-02 12:48:19 Has anyone been testing pkgsrc from netbsd on alpine? 2015-09-02 12:48:55 I would really want to swap over to alpine, but there are a few, for me, important packages missing, which I hope I would get in other ways. 2015-09-02 12:49:33 The packages I mostly miss are: thunderbird, unison and xmonad 2015-09-02 12:50:03 oh, and wxmaxima 2015-09-02 13:10:09 use any tiling wm instead of xmonad 2015-09-02 13:10:32 xmonad depends on a haskell runtime.. all haskell implementations are hopelessly broken 2015-09-02 13:12:37 nsz: Sure, I could try other wm. But I would definitely want to continue with thunderbird. It would be super much work to change to another client as I have all folders and mails from many years stored in my thunderbird profile. And wxmaxima I use quite a bit for solving difficult equations and other maths stuff. 2015-09-02 13:15:02 If you have any alternative to unison, I would be interested to try. unison does 2-way syncing. Assume I use two laptops. If I delete one file in one laptop, then sync to server, the file gets deleted from server. Next time I sync another laptop, the file also gets deleted there. 2015-09-02 13:15:07 The same with adding files 2015-09-02 13:17:07 engblom: I use git-annex to sync my files across computers 2015-09-02 13:17:25 CandyAngel: Thanks! will check up! 2015-09-02 13:17:57 Among other things.. like being the backend to my VR environment (think.. Second Life) 2015-09-02 13:18:22 The developer of git-annex is really nice too 2015-09-02 13:19:17 He implemented several things I suggested, even if they were kinda niche 2015-09-02 13:20:33 CandyAngel: How do you install it? There seem to not be any ready package for it when doing 'apk search git' 2015-09-02 13:21:23 I don't know for Alpine, I haven't used it in that yet >.< 2015-09-02 13:21:25 Sorry 2015-09-02 13:22:47 Oh, git-annex seem to be coded in Haskell. That means it will not work in alpine for the same reason xmonad does not work 2015-09-02 13:23:38 Aw :( 2015-09-02 13:23:44 That sucks for 2 reasons 2015-09-02 13:23:49 1) That means I can't use it 2015-09-02 13:24:01 2) That means you can't use it either :P 2015-09-02 13:24:12 It's err.. one of the few bits of software I *really* like 2015-09-02 13:24:42 (like, kinda fangirl over) 2015-09-02 13:26:12 Well, then the only hope for getting this stuff would be pkgsrc, if it works with alpinelinux 2015-09-02 13:26:42 Well 2015-09-02 13:26:52 There is quite a few things I want to use 2015-09-02 13:27:09 So I will probably try pkgsrcing them and see if I get anywhere 2015-09-02 13:27:45 Because I still want to use Alpine as my "everyday" VM distro 2015-09-02 13:29:04 I really dislike the way most linux distros have been going the latest time and I am still a systemd refugee, trying to get a good alternative. I love BSD but the lack of a good virtualization platform makes it not well suited for my work 2015-09-02 13:30:03 Alpine is maybe the best distro I have seen as it throws away much of the GNU-junk. 2015-09-02 13:31:09 Mhm 2015-09-02 13:34:38 I think there is going to be a combination of systemd-craziness and people wanting a systemd-free one that people will jump to the most mature one 2015-09-02 13:34:51 Even if it is, say, based on Debian but they don't like Debian 2015-09-02 13:35:08 It looks like bootstrapping pkgsrc fails on alpine :/ 2015-09-02 13:35:17 Interia is very powerful though 2015-09-02 13:35:20 Err 2015-09-02 13:35:26 Inertia* 2015-09-02 13:40:01 I reluctantly had to upgrade those distros I were using and got systemd. Since then I have had so many problems. There has been racing conditions on boot, super long boot time etc 2015-09-02 13:40:15 I would simply want a clean distro without systemd 2015-09-02 13:41:43 Yup 2015-09-02 13:42:18 And I had the same with Arch 2015-09-02 13:42:30 It was all set up just how I wanted it, worked fine, no upgrade problems 2015-09-02 13:43:01 If my audio stopped working, I'd uninstall pulseaudio (and whatever had pulled it in) and my sound would be back 2015-09-02 13:43:19 Then systemd installed and, quite literally, broke my entire setup :| 2015-09-02 13:43:34 Personally, I never liked Arch. They are too eager to rush on everything new. I like the concept of rolling distros, but Arch was not what I wanted. 2015-09-02 13:44:20 Fair enough 2015-09-02 13:44:55 I'm probably going to try Gentoo next :P 2015-09-02 13:45:03 I used openbsd-current as a "rolling distro" for many, many years. But the need for virtualization drove me to use Linux. 2015-09-02 13:45:05 Fedora -> Ubuntu -> Arch -> Gentoo 2015-09-02 13:45:14 After Gentoo.. LFS? :D 2015-09-02 13:45:30 OpenBSD is nice as long as you have a TPx220 heh 2015-09-02 13:45:34 I tried Gentoo, but the super long compilation times for the upgrades killed the joy of using it. 2015-09-02 13:46:02 Hmm 2015-09-02 13:46:05 I have once done a LFS, and that was enough. Way too much work 2015-09-02 13:46:05 I expect hardware support and power management to be sub-par if compared with linux 2015-09-02 13:46:34 I don't think I will mind compile times, providing I don't have to invest a lot of *my* time to get what I want 2015-09-02 13:46:40 If the computer is doing it, that's fine :P 2015-09-02 13:47:00 TheLemonMan: Not much anymore. WiFi support might even be better for OpenBSD. Nowadays OpenBSD supports all stuff like frequency scaling, sleep/resume etc 2015-09-02 13:47:00 engblom: i use syncthing 2015-09-02 13:47:49 engblom, good to know! did they get uefi/gpt support ? 2015-09-02 13:47:52 [TheLemonMan] engblom, good to know! did they get uefi/gpt support ? 2015-09-02 13:48:06 w..what ? 2015-09-02 13:48:40 TheLemonMan: they still use MBR, but you can have use bigger disks and partitions than supported by MBR. I have done that. 2015-09-02 13:49:08 ncopa: Thanks, syncthing looks from a first glance to be OK for my tasks. 2015-09-02 13:49:30 it works okish i think 2015-09-02 13:49:35 engblom, https://github.com/yasuoka/openbsd-uefi heheh 2015-09-02 13:49:57 Oh, before I forget (though I think I could guess correctly) 2015-09-02 13:50:19 engblom: and i use claws-mail instead of thunderbird 2015-09-02 13:50:21 Is this chat like.. strictly Alpine only? (seems very stupid considering the current discussion, but there you go) 2015-09-02 13:50:41 but i suppose we need add thunderbird too 2015-09-02 13:50:44 ncopa: claws-mail is not able to import a thunderbird profile? 2015-09-02 13:50:51 dunno 2015-09-02 13:51:12 we should probably add thunderbird 2015-09-02 13:51:20 just been so crazy busy lately 2015-09-02 13:51:22 [ncopa] just been so crazy busy lately 2015-09-02 13:51:47 hey strongid 2015-09-02 13:52:17 strongid: whats up? 2015-09-02 13:53:39 ncopa, I suspect it's a bot 2015-09-02 13:54:07 seems so yes 2015-09-02 13:56:36 ncopa: Do you see any chance to get wxmaxima? It is made using common lisp... 2015-09-02 13:57:39 dunno 2015-09-02 13:57:48 its likely possible 2015-09-02 14:03:26 wxmaxima requires SBCL for compiling... 2015-09-02 14:03:30 [algitbot (guest)] http://xkcd.com/303 2015-09-02 14:04:49 https://andrejv.github.io/wxmaxima/develop.html 2015-09-02 14:29:02 is there a way to add files to alpine-iso so that they get included in the make PROFILE=myiso iso? 2015-09-02 14:33:32 so that i can end up with apks/ boot/ and files/ 2015-09-02 14:36:38 and once i boot the cdrom they are available in /media/cdrom/files 2015-09-02 14:43:57 well, creating a basic iso, and then playing with the tmp dir that make create, then running make again seems to work 2015-09-02 14:55:23 ScrumpyJack: you still interested in the package cache stuff? 2015-09-02 14:58:39 yup 2015-09-02 14:59:34 trying to build lxc containers in a host where the host pull the packages from cache (not net) 2015-09-02 14:59:38 https://github.com/Orochimarufan/lxc-tools/blob/master/share/templates/lxc-alpine2#L43 2015-09-02 15:00:34 it's this line "$APK add -U --initdb --root $rootfs $opt_arch "$@" alpine-base" that needs tweeking 2015-09-02 15:00:41 nope 2015-09-02 15:00:49 ACTION is thinking 2015-09-02 15:01:05 you need to link your cache to $rootfs/etc/apk/cache is all 2015-09-02 15:01:55 In my case, I have lxc mount the global cache into the container at $rootfs/var/cache/apk and link that to $rootfs/etc/apk/cache 2015-09-02 15:02:01 https://github.com/Orochimarufan/lxc-tools/blob/master/share/templates/lxc-alpine2#L42 2015-09-02 15:02:36 ok, this is for when you run apk inside the lxc right? 2015-09-02 15:02:50 (Note that that template is heavily modified from the stock lxc-alpine) 2015-09-02 15:02:52 [Orochimarufan (guest)] (Note that that template is heavily modified from the stock lxc-alpine) 2015-09-02 15:03:08 wtf was that 2015-09-02 15:03:25 (moo) 2015-09-02 15:03:32 (Note moo) 2015-09-02 15:04:19 Anyway, AFAIK, "apk add --root $root" uses $root/etc/apk/cache for the cache 2015-09-02 15:05:49 Just make sure that $rootfs/etc/apk/cache points to your cache and everything should work out 2015-09-02 15:06:34 You obviously want to put an additional mount into the container so running apk from inside the container later also uses the cache 2015-09-02 15:07:13 (Unless you're security focused: I'm not sure how it's handled by apk but a shared cache COULD be a security issue) 2015-09-02 15:09:55 --root is the DESTDIR, what has that got to do with the location of the cache? 2015-09-02 15:10:16 does apk search for the cache in --root? 2015-09-02 15:10:21 I think so 2015-09-02 15:11:15 it would be good to have a apk --root=/path/to/chroot --cache=/mycache 2015-09-02 15:13:53 "apk" was a poor choice of wording... all you get with google is Android's APKs :( 2015-09-02 15:16:14 ln -s ../../var/cache/apk "$rootfs/etc/apk/cache" worked, thanks 2015-09-02 15:16:35 now building 120 alpine lxc containers is much quicker :) 2015-09-02 15:16:37 only if your cache is actually at $rootfs/var/cache/apk, but yeah 2015-09-02 15:16:49 why so many? :O 2015-09-02 15:16:54 tests 2015-09-02 15:17:00 let me time it 2015-09-02 15:17:49 ScrumpyJack: I like your style 2015-09-02 15:18:08 real 0m 43.85s 2015-09-02 15:18:18 "I can make 120 Alpine LXC containers... *pauses* .. lets do that" 2015-09-02 15:18:47 heh 2015-09-02 15:18:56 that would take a while w/o a cache :P 2015-09-02 15:19:52 Wha the heck are there patches in aports/main/abuild? abuild is an Alpine project :O 2015-09-02 15:20:39 I guess you're keeping the "maintenance branch" as patches instead of an actual branch? :O 2015-09-02 15:21:04 About an hour until I do battle with pci-stub again :| 2015-09-02 15:21:11 heh 2015-09-02 15:21:15 output of script recording 2015-09-02 15:21:17 http://st.ilet.to/lxc-show 2015-09-02 15:21:46 ... 2015-09-02 15:21:56 If I don't have it working by.. hmm... 8pm, I'm just going to blacklist nouveau and try to unbind snd_hda_intel manually 2015-09-02 15:22:20 what's lxc-show? 2015-09-02 15:22:44 is the output of a script recording 2015-09-02 15:23:07 anyway: 2015-09-02 15:23:11 root@Toukyou ~ # lxc-ls 2015-09-02 15:23:11 Akihabara Ebisu Ikebukuro IkebukuroDeb Shibuya Shinagawa Shinjuku 2015-09-02 15:23:17 see what I did there? :P 2015-09-02 15:24:07 /usr/bin/script is owned by util-linux-2.26.2-r0 2015-09-02 15:24:34 well yeah, but I don't have a lxc-show command 2015-09-02 15:24:48 ah, that's just a shell script 2015-09-02 15:25:01 Not that it seems particularly useful ;P 2015-09-02 15:25:16 for i in `sudo lxc-ls`;do echo; sudo lxc-info -n $i ;done 2015-09-02 15:25:37 I thought something along those lines 2015-09-02 15:25:59 not very useful for 120 lxc, no 2015-09-02 15:26:08 I wouldn't use sudo in scripts though^^ rather, run the script in sudo instead 2015-09-02 15:26:40 hmm I got an idea 2015-09-02 15:26:47 this is my work desktop, i'm lazy and slack 2015-09-02 15:28:38 http://st.ilet.to/lxc-running 2015-09-02 15:29:23 neat 2015-09-02 15:29:29 I usually just us lxc-ls --fancy 2015-09-02 15:30:04 .. these numbers aren't 0 padded :| 2015-09-02 15:30:07 The sortinggggggggg 2015-09-02 15:30:09 :P 2015-09-02 15:30:22 heh 2015-09-02 15:30:36 no lxc-ls --fancy on alpine linux 2015-09-02 15:31:02 why's that? 2015-09-02 15:31:16 what's your lxc version :O 2015-09-02 15:32:47 1.1.3 2015-09-02 15:34:00 bah, i forgot how to use scriptreplay 2015-09-02 15:41:21 Hm 2015-09-02 15:41:33 How easy is it to recompile the kernel to make a module a builtin? 2015-09-02 15:41:41 If that is the only change I want to make 2015-09-02 15:42:29 the kernel used in alpine linux? 2015-09-02 15:42:37 Yeah 2015-09-02 15:43:36 in that case i would get a alpine linux build environment together and play with the aports tree 2015-09-02 15:44:17 why do you need a a custom kernel? 2015-09-02 15:44:45 I want to see if putting pci-stub in the kernel will make it get assigned to the devices I want to passthrough to the VMs 2015-09-02 15:45:08 Not very confident (dmesg doesn't show pci-stub even trying to add them) but maybe it will work.. 2015-09-02 15:45:38 Like 2015-09-02 15:45:58 Maybe nouveau and snd_hda_intel are being loaded first, because pci-stub isn't a builtin 2015-09-02 15:46:03 Is my thinking anyway 2015-09-02 15:46:32 CandyAngel: building a vanilla kernel with pci-stub built in now 2015-09-02 15:47:28 it might take a while, build box is not very beefy 2015-09-02 15:47:33 Oh, thank you! 2015-09-02 15:47:56 Trust me, I appreciate the help regardless of how long it takes >.< 2015-09-02 15:48:24 I've pretty much run out of ideas at this point 2015-09-02 15:48:56 Something is working, because the modules does end up loaded (shows in lsmod after boot) 2015-09-02 15:49:15 But no logs saying it tried to get anything in dmesg 2015-09-02 15:49:50 When I would be expecting to see something like.. 2015-09-02 15:49:58 pci-stub: add 1002:6719 sub=FFFFFFFF:FFFFFFFF cls=00000000/00000000 2015-09-02 15:50:27 Hm 2015-09-02 15:50:50 I wonder.. if I unbind the drivers so they are unclaimed, then run: modprobe pci-stub ids=aaaa:bbbb 2015-09-02 15:51:11 It should try to get it and give me dmesg logs..? 2015-09-02 15:51:25 what does readlink say? 2015-09-02 15:51:38 readlink? On what? 2015-09-02 15:51:38 readlink /sys/bus/pci/devices/... 2015-09-02 15:51:56 I'm not sure 2015-09-02 15:52:02 I find out what drivers are in use with umm 2015-09-02 15:52:10 lspci -nnv -s 1: 2015-09-02 15:52:18 And it says it on the bottom line 2015-09-02 15:52:39 Kernel driver in use: snd_hda_intel 2015-09-02 15:52:44 For example 2015-09-02 15:52:51 this is for your graphics card passthrough? 2015-09-02 15:52:54 Mhm 2015-09-02 15:52:57 ScrumpyJack: This is how I roll: http://oro.sodimm.me/~hinata/misc/lxc-lsc.htm :P 2015-09-02 15:53:24 pci-stub is supposed to be assigned to those devices early, so the proper drivers don't get assigned to them 2015-09-02 15:53:48 And then you pass them to the VM 2015-09-02 15:54:01 CandyAngel: The problem with GPUs is that some don't like being initialized multiple times IIRC 2015-09-02 15:54:13 what's the device id? 2015-09-02 15:54:30 Orochimarufan: Yeah, I'm aware of that 2015-09-02 15:55:18 ScrumpyJack: 10de:1086 (nVidia VGA) and 10de:0e09 (nVidia audio) 2015-09-02 15:55:32 On my computer 2015-09-02 15:56:21 is that was lspci says? 2015-09-02 15:56:25 Mhm 2015-09-02 15:56:36 That's my GTX 570 2015-09-02 15:57:13 can you show us ls /sys/bus/pci/devices/ 2015-09-02 15:57:35 Orochimarufan: pretty :) 2015-09-02 15:57:41 Indeed :) 2015-09-02 15:57:53 ACTION loves colors 2015-09-02 15:58:31 care to share? 2015-09-02 15:58:40 uh sure 2015-09-02 15:59:00 THis? http://pastebin.archlinux.fr/1510639 2015-09-02 15:59:10 And I have no idea why it is a french pastebin :P 2015-09-02 15:59:53 ScrumpyJack: https://github.com/Orochimarufan/lxc-tools/blob/master/bin/lxc-lsc There you go 2015-09-02 16:00:04 yes. which is your card? 2015-09-02 16:00:24 Ohhhh 2015-09-02 16:00:40 You meant PCI device ID, not vendor/device device id 2015-09-02 16:00:42 Sorry 2015-09-02 16:01:02 01:00.0 and 01:00.1 (vga/audio) 2015-09-02 16:01:16 So.. line 18 and 19 2015-09-02 16:01:41 what does readlink /sys/bus/pci/devices/0000\:01:00.0 2015-09-02 16:01:49 does it link to stub? 2015-09-02 16:02:34 No, it goes to ../../../devices/pci0000:00/0000:00:01.0/0000:01:00.0 2015-09-02 16:02:47 so it's defo not under stub control 2015-09-02 16:05:09 Hm 2015-09-02 16:05:11 is qemu+kvm? 2015-09-02 16:05:15 or xen? 2015-09-02 16:05:19 I'll be back in a moment, I have to head home 2015-09-02 16:05:26 It's for QEMU/KVM 2015-09-02 16:05:31 You use pci-stub on the host 2015-09-02 16:05:49 To prevent it from "taking" devices intended for the guests 2015-09-02 16:15:49 Sorry if I am explaining stuff you already know. i dont mean to patronise or anything 2015-09-02 16:16:16 (on phone walking home :P) 2015-09-02 16:16:45 i'm just looking at qemu/kvm docs 2015-09-02 16:17:28 you can modprobe pci_stub, and does it create a tree in /sys/bus/pci/drivers/pci-stub/ ? 2015-09-02 16:21:45 i got it to work with alpine linux 3.18.20-1-grsec 2015-09-02 16:22:09 for my audio device i did this 2015-09-02 16:25:05 http://st.ilet.to/stub 2015-09-02 16:30:22 Yeaj, thats the manually unbinding method 2015-09-02 16:30:59 But the friver may have already initialised it at that point 2015-09-02 16:31:25 So then it might not work in he VM 2015-09-02 16:34:04 but once you unbind it and bind it to stub, you can assign it to the guest no? 2015-09-02 16:37:22 hello! has alpine qemu on repositories and kvm module available? thanks 2015-09-02 16:37:23 you're unbinding it from whatever driver had got hold of it 2015-09-02 16:37:36 lupyxe: yup 2015-09-02 16:37:53 version 2.4 2015-09-02 16:38:13 thanks ScrumpyJack! 2015-09-02 16:39:09 there is also an alpine-xen build that you might like to try. gives you alpine linux as a dom0 for xen 2015-09-02 16:39:31 ScrumpyJack: yea, but GPUs sometimes break when initialized twice 2015-09-02 16:39:52 So you need to prevent the host OS from initializing it to let the guest os driver do it 2015-09-02 16:40:16 does this one break when grabed by the nvidia driver? 2015-09-02 16:40:22 yeah, i know, ScrumpyJack, i only want to have other vm available too :) 2015-09-02 16:40:22 grabbed even 2015-09-02 16:42:41 i'm sure you've tried putting the modules in /etc/modprobe.d/blacklist.conf, but it's worth mentioning 2015-09-02 16:42:44 I dont think it is ever grabbed by anything other than the nvidia driver as you cant modprove nvidia after booing 2015-09-02 16:43:17 I cant blacklist snd_hda_intel because the host audio uses that 2015-09-02 16:43:24 booting* 2015-09-02 16:43:48 Blacklisting nouveau is fine 2015-09-02 16:45:06 so you tried booting, rmmod nouveau and any other relevant modules, then modprobe pci_stub and manually unbinding/binding to stub? 2015-09-02 16:46:00 (your kernel is still building) 2015-09-02 16:55:21 gtg 2015-09-02 16:58:29 what's the difference between the grsec and pae kernels? 2015-09-02 17:03:39 grsec probably has more security related patches 2015-09-02 17:04:12 PAE is pretty much standard at this point 2015-09-02 17:04:17 so the pae kernel desn't have the grsec? 2015-09-02 17:04:31 I thought PAE was built into the kernel and couldn't even be disabled at this point 2015-09-02 17:04:53 grsec probably has PAE enabled 2015-09-02 17:04:57 ah ok 2015-09-02 17:05:05 so just normal kernel and grsec kernel then 2015-09-02 17:05:17 these days, its rare to find either kernels OR hardware WITHOUT PAE 2015-09-02 17:06:01 I guess it's still more likely in Alpine than in most other distros as Alpine is often used on routers and other devices like that *shrug* 2015-09-02 17:08:57 considering PAE is supported on all Pentium Pros and above. except the Pentium M and all Athlons and above. 2015-09-02 17:09:04 yah... 2015-09-02 17:09:58 embedded CPUs are another thing though, I know we have some VIA thin-clients that don't support PAe cause they don't conform to x86 architecure 100% 2015-09-02 17:15:01 yeah 2015-09-02 17:15:17 You'd have to ask ncopa to be certain 2015-09-02 19:18:48 So apparently abuild now rejects uncompressed man pages? 2015-09-02 19:19:08 yeah, seems so. 2015-09-02 19:19:15 and it does not compress them expect for -doc 2015-09-02 19:19:18 annoying 2015-09-02 19:20:24 uhm... it doesn't seem to do it for -doc either 2015-09-02 19:21:00 >>> ntp*: Running postcheck for ntp-doc 2015-09-02 19:21:00 >>> ERROR: ntp*: Found uncompressed man pages: 2015-09-02 19:21:00 /usr/share/man/man1/sntp.1 [...] 2015-09-02 19:21:18 huh. that's strange 2015-09-02 19:21:20 is there some option I need to set? 2015-09-02 19:21:24 i don't think so 2015-09-02 19:21:29 maybe it's missing some paths then 2015-09-02 19:21:31 sounds abuild bug 2015-09-02 19:27:10 >>> v4l-utils*: Running postcheck for qv4l2 2015-09-02 19:27:10 >>> WARNING: v4l-utils*: Found /usr/share/man but package name doesn't end with -doc 2015-09-02 19:27:10 >>> ERROR: v4l-utils*: Found uncompressed man pages: 2015-09-02 19:27:17 bah 2015-09-02 19:28:53 Quick Fix (TM): 2015-09-02 19:28:59 for manfile in "$subpkgdir"/usr/share/man/*/*; do 2015-09-02 19:28:59 gzip "$manfile" 2015-09-02 19:28:59 done 2015-09-02 20:25:07 -.- 2015-09-02 20:25:21 Every time I follow these instructions, something doesn't work 2015-09-02 20:28:34 Oh! 2015-09-02 20:28:37 Argh 2015-09-02 20:28:39 Never mind, my error 2015-09-02 20:29:26 I keep forgetting I've changed the primary display device to PCI-E for Arch 2015-09-02 20:34:45 Any idea why the alpine logo doesn't display on my custom-aports README on github anymore? 2015-09-02 20:44:30 Hmm it seems to have fixed itself 2015-09-02 20:54:54 Okay 2015-09-02 20:55:26 pci_stub now attempts to stub devices! 2015-09-02 20:55:37 But.. I guess it loads too late to actually succeed :/ 2015-09-02 21:01:31 blacklist the real modules? 2015-09-02 21:06:56 I can blacklist nouveau, because my hosts GPU doesn't use it 2015-09-02 21:07:06 But my host audio uses the same module as the nvidia cards do.. 2015-09-02 21:07:12 Also 2015-09-02 21:07:32 I just checked modules.order and pci_stub is listed before (252) nouveau (436) 2015-09-02 21:10:10 So currently, I have pci_stub in /etc/modules and the ids in /etc/modprobe.d/vfio.conf 2015-09-02 21:10:17 And it appears in dmesg now 2015-09-02 21:10:42 I've also tried adding it to the modules= list on the cmdline and adding pci_stub.ids= too 2015-09-02 21:10:57 Still isn't getting the devices first 2015-09-02 21:12:12 hmm no idea 2015-09-02 21:13:37 ! 2015-09-02 21:13:40 So 2015-09-02 21:14:17 I got pci_stub to grab it instead of nouveau! 2015-09-02 21:14:31 but snd_hda_intel is still getting the audio device of the VGA card 2015-09-02 21:14:38 I specified 'nomodeset' on the cmdline 2015-09-02 21:14:53 hmm 2015-09-02 21:15:28 Basically, in dmesg 2015-09-02 21:15:44 pci_stub is being loaded at 11(ms?) in 2015-09-02 21:15:54 Personally, I'd just use a container for the linux guest. that way you don't have to pass anything 2015-09-02 21:16:03 But snd_hda_intel is loaded at 9ms 2015-09-02 21:16:12 Thought that still leaves the windows guest hmm 2015-09-02 21:16:26 Yeah, I still need it for Windows either way 2015-09-02 21:16:52 plus, I won't be using Alpine for my main linux VM because it is missing sooo much software that I use 2015-09-02 21:17:28 Somehow, I just need to get pci_stub to load earlier 2015-09-02 21:17:44 That's why I was asking about recompiling the kernel with it as a builtin 2015-09-02 21:18:26 I wish i could do that at work algitbot, but alas, I'm not actually even supposed to be coding and when i do anyway.. it's in perl :P 2015-09-02 21:18:32 (and yes, I know its a bot) 2015-09-02 21:19:53 CandyAngel: what does your /etc/modules contain? 2015-09-02 21:21:12 Am I allowed to just type it here or should I pastebin it? 2015-09-02 21:21:32 how much is it? 2015-09-02 21:21:57 7 lines 2015-09-02 21:23:45 you should be fine replacing newlines with spaces or something 2015-09-02 21:23:53 Oh yeah, good point :P 2015-09-02 21:24:16 af_packet, raid1, psmouse, i915, fbcon, kvm-intel, pci_stub 2015-09-02 21:24:54 hmm doesn't look like a problem 2015-09-02 21:25:53 I wonder if you can blacklist the audio and then un-blacklist it using an init script after pci_stub has grabbed the right devices (smells like old fish, but it might work^^) 2015-09-02 21:42:19 Unblacklist? 2015-09-02 21:49:47 Okay 2015-09-02 21:49:50 I see a problem here 2015-09-02 21:49:57 pci_stub isn't in my initfs 2015-09-02 21:50:32 That explains why it is not being loaded early! 2015-09-02 21:54:15 there you go 2015-09-02 21:55:55 But 2015-09-02 21:55:59 I don't know why it isn't there 2015-09-02 21:56:04 It's in /etc/modules 2015-09-02 21:56:10 rebuild it? 2015-09-02 21:56:26 mkinitfs -l is what is telling me it isn't there 2015-09-02 21:56:48 So it isn't going to add it if I rebuild it (and I rebuilt it after adding to /etc/modules anyway) 2015-09-02 21:57:00 Because I thought /etc/modules was how you got it into initfs :P 2015-09-02 21:57:32 its how it works on debian anyway 2015-09-02 21:58:11 I did umm 2015-09-02 21:58:18 mkinitfs -l | grep stub 2015-09-02 21:58:34 maybe look into /etc/mkinitfs/ ? 2015-09-02 21:58:43 And it only show i2c-stub 2015-09-02 22:00:18 It looks like if I write a features file for it, it'll get included 2015-09-02 22:00:31 (read the source for mkinitfs) 2015-09-02 22:01:15 I think /etc/modules is just used by umm.. /etc/init.d/modules 2015-09-02 22:01:36 Hm, maybes not 2015-09-02 22:02:21 Oh yeah, it is 2015-09-02 22:02:35 Okay.. one features file, coming up! 2015-09-02 22:10:06 YES 2015-09-02 22:10:09 I did it! 2015-09-02 22:10:51 congrats :D 2015-09-02 22:10:52 I wrote a feature file that adds in pci_stub to initfs 2015-09-02 22:11:03 And added pci_stub to modules= on cmdline 2015-09-02 22:11:09 And they are now stubbed 2015-09-02 22:11:13 ACTION does a little dance 2015-09-02 22:11:16 nice 2015-09-02 22:13:15 I wonder if that is the sort of thing that might be accepted upstream as a patch.. 2015-09-02 22:13:27 Or if it would be preferred to be kept minimal 2015-09-02 22:48:08 ACTION is back in her Alpine guest now \o/ 2015-09-02 22:52:54 Yay! 2015-09-02 23:00:42 Okay 2015-09-02 23:00:58 let's see if I can install say.. Mint, with passthrough 2015-09-03 05:55:25 good morning! 2015-09-03 05:56:20 I'd like to know if there's some progress on opensmtpd on musl. 2015-09-03 05:56:59 there's http://bugs.alpinelinux.org/issues/4012 for reference. 2015-09-03 06:05:54 Gottox, is opensmtpd apkbuild somewhere; the bug tracker seems to have only libasr there 2015-09-03 06:14:32 Gottox, seems libasr wants res_randomid() which is not present onmusl 2015-09-03 06:54:01 CandyAngel: here is your kernel build http://st.ilet.to/linux-vanilla-3.18.20-r0.apk 2015-09-03 07:02:52 ScrumpyJack: Thankies ^_^ 2015-09-03 07:09:38 fabled: nice spot. annoying that libasr builds fine without res_randomid() 2015-09-03 07:14:46 so no chromium or google chrome under alpine? 2015-09-03 07:14:59 fabled: thanks for that info! 2015-09-03 07:15:01 there is one in testing i think 2015-09-03 07:15:12 It's annoying that res_randomid is not very random .. 2015-09-03 07:15:20 it feels crazy that this would be the issue to migrate my workstation to alpine :D 2015-09-03 07:15:29 return 0xffff & __getpid(); 2015-09-03 07:15:32 Gottox: https://github.com/OpenSMTPD/libasr/issues/8 2015-09-03 07:16:06 pcarrier: perhaps the issue is that you favour a browser over a distribution ;) 2015-09-03 07:16:07 ah 2015-09-03 07:16:08 btw, you guys did an awesome job! ;) 2015-09-03 07:16:15 -did+do 2015-09-03 07:16:49 ScrumpyJack: perhaps, but you might be missing how important the difference is, when you spend the most part of your day working in a company that built on web 2015-09-03 07:17:02 just sayin'. 2015-09-03 07:17:29 cause however great the firefox experience might be in that field 2015-09-03 07:17:47 the learning investment in something like the chrome developer tools is a bit more than "oh that menu is a bit different" 2015-09-03 07:17:50 pcarrier: i was being facetious, i did't deserve a response :) 2015-09-03 07:18:16 fair enough, but I feel like that's a general feeling people have about browsers 2015-09-03 07:18:38 even when they're completely OK trolling on text editors for longer than I ever cared for 2015-09-03 07:18:52 you are free to use what you like - i have no quarms about it :) 2015-09-03 07:19:03 Everyone knows bwan is the best text editor 2015-09-03 07:19:22 give chrome in testing a try, it would help the packager 2015-09-03 07:19:23 (Butterfly Wings and Needles) 2015-09-03 07:19:50 oh yeah no worries ScrumpyJack, I was just going on about the topic, didn't take you seriously 2015-09-03 07:20:15 cool :) 2015-09-03 07:20:20 wait how did I miss that? :D 2015-09-03 07:21:03 ScrumpyJack: where's that "chrome in testing" you're referring to? 2015-09-03 07:21:11 The likelihood of overlooking something is proportional to how silly you look overlooking it and how obvious it is? 2015-09-03 07:21:11 that's a source package somehow? 2015-09-03 07:21:19 i've just looked and it's not there 2015-09-03 07:21:24 riiight. ok :D 2015-09-03 07:21:43 to be clear, using testing wouldn't have been an issue to migrate my workstation to alpine :D 2015-09-03 07:22:18 the part that saddens me is that if I don't use alpine 2015-09-03 07:22:25 the most viable alternative right now is basically fedora 2015-09-03 07:22:28 I think the clandmeter built it to test it pushed it out somewhere 2015-09-03 07:22:46 in terms of benefits over investment, because I don't feel passionate about anything else 2015-09-03 07:23:16 let me see if i can find a link 2015-09-03 07:23:17 ScrumpyJack: [Citation needed] [No seriously I need this] 2015-09-03 07:23:38 and I should totally be willing to put in the effort 2015-09-03 07:24:28 but omg I do not want to fight with the chromium build system, nor weird kerberos libraries crap in the case of google chrome wrapped in a mini-glibc-based-set-of-system-libraries :D 2015-09-03 07:25:47 my personal laptop and main home device: a chromebook. 2015-09-03 07:27:02 OK is it crazy to wish apk was renamed to avoid the conflict with android when googling? :D 2015-09-03 07:27:22 that feels like an extreme measure, but my oh my it's annoying 2015-09-03 07:29:16 yup, fedora "just makes sense". it's that or debian/ubuntu/fedora/opensuse to get google chrome packages with security updates I can trust 2015-09-03 07:29:45 it's in edge chromium-44.0.2403.157-r0 2015-09-03 07:29:58 maybe I could live in an alpine container for most things 2015-09-03 07:30:05 only on x86_64 2015-09-03 07:30:31 yeah I don't touch anything else yet. looking forward to arm being viable enough though 2015-09-03 07:30:56 ScrumpyJack: huuu. so I'd just miss the flash player. 2015-09-03 07:31:14 that's pretty damn viable. 2015-09-03 07:33:52 lvm and luks is supported, not need to destroy my system just yet. very nice 2015-09-03 08:21:26 Please point me to any information on how I should install Alpine on rpi2 2015-09-03 08:27:34 ncopa: have you guys figured out w/ to do with the grsec licensing? 2015-09-03 08:30:03 frankS2: i need write to spender/pax 2015-09-03 08:30:06 i have some ideas 2015-09-03 08:38:43 Also, as a bit of throwback to the systemd conversation yesterday, when I rebooted into Arch last night so I could connect to my computer today, systemd threw a fit and I had errors and timers on boot :| 2015-09-03 08:39:10 My Arch hasn't been changed for like.. 2 weeks (as I'm been trying to migrate to my VM setup) 2015-09-03 08:40:09 Anyone here from the Philippines? Just curious. 2015-09-03 08:40:25 Sachiru: England here 2015-09-03 08:45:51 Cool 2015-09-03 08:45:54 Anyone else? 2015-09-03 08:58:28 engblom: have a look at the first part of this http://wiki.alpinelinux.org/wiki/RPI_Video_Receiver 2015-09-03 09:11:33 ScrumpyJack: Thanks! 2015-09-03 09:14:02 By reading that page, it looks like alpine on rpi is read-only. Is there no chance to get alpine fully installed and upgradeable on rpi? 2015-09-03 09:20:42 ncopa: could you trigger a rebuild of cdw? so:libcdio.so.15 (missing 2015-09-03 09:21:09 engblom: yes 2015-09-03 09:21:20 have you managed to install and can boot? 2015-09-03 09:22:37 ScrumpyJack: I just read the wiki, I did not try it yet. But from the wiki it talked about how it is RO. Will it load to RAM? Or how will it repartion the SD card? 2015-09-03 09:29:15 it loads to ram, but you can write to the DS card 2015-09-03 09:29:51 have a look at http://wiki.alpinelinux.org/wiki/Alpine_local_backup 2015-09-03 09:31:00 for the RPi install, it's all set up already. just do stuff, and run lbu commit to write your changes to the SD card 2015-09-03 09:34:45 ScrumpyJack: What would happen if I would run setup-alpine? Could I get it to be like a normal installation without the need to run lbu? 2015-09-03 09:36:10 I planned to try to see if the official oracle java8 would run on alpine on rpi2. It is quite huge to keep in RAM. And installation requires both the tar.gz file, extracting, installing 2015-09-03 09:46:40 without lbu, not at the moment 2015-09-03 10:04:14 engblom, ScrumpyJack : it's perfectly doable to install to SD card so that SD is root. and that probably needs to be done to run openjdk 2015-09-03 10:04:33 though - i generally try to avoid that, as having root on SD usually wears out the SD sooner or later 2015-09-03 10:09:15 armhf port of alpine does not have package for openjdk. This is why I wanted to try the offical oracle version. 2015-09-03 10:14:15 While I was waiting for response here, I tested to see if I could get oracle java8 to install on amd64. It failed. Looks like it is dynamically linked to glibc... 2015-09-03 10:14:58 Unless I get java to run on rpi, there is no reason for me to even try getting alpine on rpi. 2015-09-03 10:15:43 s/I get/I find a way/ 2015-09-03 10:24:34 engblom, yes, openjdk is not present on arm. iirc, openjdk7 builds, but has TEXTRELs so the build is not accepted currently. 2015-09-03 10:24:39 i have not tried openjdk8 on arm 2015-09-03 10:24:47 because it needs openjdk7 to bootstrap 2015-09-03 10:29:36 how do you guys bootstrapped openjdk? - through oracle-jdk with glibc? 2015-09-03 10:36:12 fabled: What does TEXTRELs mean? 2015-09-03 10:46:23 Gottox, gcc-java used to build icedtea openjdk7; openjkd7 used to build openjdk8 2015-09-03 10:46:50 engblom, it's certain type of relocation we avoid to due to security reasons 2015-09-03 10:47:04 usually it's the result of using assembler in non-compatible mode 2015-09-03 10:47:39 which was true of openjdk too; it has large chunks of arm assembler, and few parts of it are incompatible in sense that they result in text relocations 2015-09-03 10:48:20 fabled: If I want to regard less of security reasons build openjdk7 for rpi, how would I proceed? Is there any ready made tree I could download/clone? 2015-09-03 10:50:40 fabled: a, so no closed source software was involved? that's awesome! 2015-09-03 10:52:04 i have a armv8 dev box for a few more days, should i try making a chrooted build (it runs shitbuntu) and hand pkgs to someone? 2015-09-03 10:53:28 Any java >= 6 would be OK for me. I do not even need a full JDK. All I need is the JRE. I am programming in Clojure and it got its own compilator running on JRE. 2015-09-03 10:59:45 Gottox, no closed source stuff, it's all built from aports. 2015-09-03 11:00:14 engblom, hmm, i think the openjdk7 apkbuild should work, but we would need to add some option to abuild to disregard TEXTRELs 2015-09-03 11:00:39 engblom, gcc-java should work for arm 2015-09-03 11:00:48 but it's not exactly full JRE environment 2015-09-03 11:01:16 gcc-java is a bit lacking behind the JRE 1.6. It does not understand all the bytecode Clojure produces 2015-09-03 11:06:58 yes, understand 2015-09-03 11:07:35 what i've seen is that other distros just disabled the TEXTREL check for arm on openjdk; if they had any policy to check that in the first place 2015-09-03 11:09:29 Is apk supporting messages during installation? An good alternative would be to show a message you need to quit from with 'q' while installing. Then the responsability is upon the one installing. 2015-09-03 11:11:56 yeah, just checked last arm buildlog of openjdk7 2015-09-03 11:11:57 >>> openjdk7*: Running postcheck for openjdk7-jre-base 2015-09-03 11:11:57 >>> ERROR: openjdk7*: Found textrels: 2015-09-03 11:11:57 TEXTREL /home/buildozer/aports/main/openjdk7/pkg/openjdk7-jre-base/usr/lib/jvm/java-1.7-openjdk/jre/lib/arm/server/libjvm.so 2015-09-03 11:11:58 >>> ERROR: openjdk7*: prepare_subpackages failed 2015-09-03 11:12:00 >>> ERROR: openjdk7: all failed 2015-09-03 11:12:02 >>> openjdk7: Uninstalling dependencies... 2015-09-03 11:25:38 fabled: where you doing something with cciss drivers? 2015-09-03 11:26:53 ScrumpyJack, not really. i have some server somewhere with cciss. but i have not been doing anything actively with that. 2015-09-03 11:26:59 ScrumpyJack, maybe it was nenolod ? 2015-09-03 11:27:03 err... kaniini 2015-09-03 11:35:25 ncopa, seems openjdk7 is difficult to fix. there's 7300 lines of assembly, with lot of offending code producing textrels 2015-09-03 11:35:34 i wonder if we should just allow textrels for that 2015-09-03 11:38:17 hum 2015-09-03 11:38:25 i suppose thats ok 2015-09-03 11:38:51 Nice! 2015-09-03 11:38:55 i doubt it is worth fixing 7k lines assembly code 2015-09-03 11:39:43 also, openjdk7 is only for bootstrapping openjdk8 2015-09-03 11:40:38 openjdk8 might have same issues 2015-09-03 11:41:16 Probably it has... It maybe would be good to show a message during installation to warn the user 2015-09-03 11:41:44 In my case (controlling some stuff through GPIO) I do not care much about security... 2015-09-03 11:42:14 the biggest issue with TEXTRELs is that it needs to allocate separate memory pages for each running process sharing the same library 2015-09-03 11:42:26 in java case it'd make it unfeasible to run multiple JVM processes 2015-09-03 11:44:43 A slightly problematic JVM is better than no JVM. My opinion is that a frank message about the state of the package would make everything OK. Then the user understands that there might be problem and he will not feel tricked. 2015-09-03 11:47:49 ah 2015-09-03 11:47:57 we even have option for skipping textrel check 2015-09-03 11:48:00 i'll apply that then 2015-09-03 11:48:16 our arm builder is quite busy, and building java takes ages. so it's gonna take a while regardless 2015-09-03 11:48:54 Ok! That is understandable. 2015-09-03 11:49:23 How soon do you think openjdk will appear in edge testing for armhf? 2015-09-03 11:49:38 Do we talk about weeks or months? 2015-09-03 11:50:25 day or two probably 2015-09-03 11:50:29 Nice! 2015-09-03 11:50:31 ncopa, http://sprunge.us/PdHV 2015-09-03 11:50:36 would ^^ be ok? 2015-09-03 12:00:59 fabled: sounds good to me 2015-09-03 12:01:02 looks good 2015-09-03 12:01:10 better than i hoped :) 2015-09-03 12:07:46 This gave me enough hope for openjdk that I am going to try installing alpine on my rpi2. 2015-09-03 12:11:45 Btw, I found a bug in edge for amd64: When upgrading from stable to edge, eth0 is lost. 2015-09-03 12:11:55 eth0 is not even visible in /proc/net/dev 2015-09-03 12:21:11 When doing setup-alpine on rpi, it does not ask me for doing a sys install. It only asks where I want the config stored. 2015-09-03 12:21:18 setup-disk says I have no disks 2015-09-03 12:22:16 How should I do to install with the SD card as root? 2015-09-03 12:25:01 Should I have a FAT32 /boot for getting rpi2 to boot? 2015-09-03 12:25:14 just one partition 2015-09-03 12:25:56 bootable flag needs to be set 2015-09-03 12:26:11 ScrumpyJack: Does rpi boot from ext4 directly? 2015-09-03 12:26:32 It looks like other distros are using fat32 for booting 2015-09-03 12:26:43 nope, the fs needs to be fat 2015-09-03 12:27:13 Rasbian got FAT32 /boot and Ext4 / 2015-09-03 12:27:14 the partition type needs to be FAT32 and the filesystem needs to be fat 2015-09-03 13:21:04 how would i go about installing AL on a HP SAS controller raid disk? 2015-09-03 13:35:49 anyone experienced with the Makefile of alpine-iso? 2015-09-03 13:36:10 we should probably rewrite that 2015-09-03 13:36:17 use update-kernel and other script to do it 2015-09-03 13:36:22 I'm trying to add a openrc init script to a standard iso 2015-09-03 13:36:38 and add hooks so we can build easily customized images w/ additional data, scripts, etc. 2015-09-03 13:38:26 build with an addition script is what I'm trying to do :) 2015-09-03 13:41:08 i mean an additional init script (in local.d) 2015-09-03 15:02:21 I've discovered a flaw in my plane 2015-09-03 15:02:30 And my plan 2015-09-03 15:03:01 .. Alpine doesn't have synergy 2015-09-03 15:28:45 Not sure what to do about that, it kinda ruins the plan 2015-09-03 15:28:57 I'll have to see if I can APKBUILD it I guess..? 2015-09-03 15:32:23 :) 2015-09-03 15:35:23 It has no GUI or anything so it shouldn't have too many dependencies.. (famous last words) 2015-09-03 15:35:53 this should help http://tinyurl.com/pmd7zyr 2015-09-03 15:36:41 Thankies 2015-09-03 15:37:13 I've had a look at the requirements and nothing jumps out as something Alpine won't have 2015-09-03 15:37:57 Python, Xtsts, Qt, libCURL and SSL 2015-09-03 15:38:00 Pretty much 2015-09-03 15:38:50 If it works, can I like.. submit it to edge/testing or something? 2015-09-03 15:39:03 build it without qt if you can 2015-09-03 15:39:39 yes, you can submit a git patch to the alpine-aports mailing list 2015-09-03 15:40:20 Coolies 2015-09-03 15:40:29 Hm 2015-09-03 15:41:00 I don't actually know why it would need Qt anyway 2015-09-03 15:41:10 There is a GUI for it but it was separate last time I used it 2015-09-03 15:41:12 configuration gui? 2015-09-03 15:41:54 Yeah, it isn't needed though 2015-09-03 15:42:00 It is configure by a plaintext file 2015-09-03 15:42:04 configured* 2015-09-03 15:43:02 I think syntool might be wanting it, that doesn't sound familiar 2015-09-03 15:43:11 have there been any updates to the opensmtpd issue? 2015-09-03 15:43:25 I noticed that libasr is available in testing 2015-09-03 15:43:27 I thought synergy went the RHEL route and didn't offer free (gratis) downloads anymore (while still being GPL2) but it looks like all the code is on github? 2015-09-03 15:44:28 Oh.. what 2015-09-03 15:44:30 Maybe 2015-09-03 15:44:34 It's been ages since I used it :/ 2015-09-03 15:45:40 Yeah 2015-09-03 15:45:44 It's free to compile it yourself 2015-09-03 15:46:15 Also, I don't trust companies who say "buy it once, get lifetime access" 2015-09-03 15:46:15 Well, the code is still up on github :), from the synergy webpage they have downloads that need an account with them. 2015-09-03 15:46:50 Got burned by Sandboxie later turning around and saying "Oh, only up to v3, then you have to pay monthly now" 2015-09-03 15:47:35 The "lifetime" refers to the lifetime of the free download program ;) 2015-09-03 15:47:58 As far as I remember, it explicitly said "all updates" 2015-09-03 15:48:17 Plus, they work around it by going "for this major version" and then doing a major version bump for.. no reason 2015-09-03 15:50:14 I dunno, maybe I'm just not a jerk and when I said "buy a license for X and get lifetime access", that means lifetime access to X 2015-09-03 15:50:37 Like.. as long as X exists, regardless of what it is called, you have a license for it 2015-09-03 15:51:46 icicled: not at the moment, but feel free to make some noise on #opensmtpd and #musl 2015-09-03 15:52:08 the more pipe up, the better 2015-09-03 15:52:10 will do 2015-09-03 15:52:32 I noticed that the related bugs upstream haven't been looked at in a long time 2015-09-03 15:54:28 well, fabled mentioned that libasr calls res_randomid() which isn't in musl 2015-09-03 15:56:02 yes, there was a patch floating around that copied the implementation from gnulibc 2015-09-03 15:56:06 it was quite silly 2015-09-03 15:56:46 i think this is part of resolv.h, and musl philosphize away from it 2015-09-03 15:56:47 ^ understatement 2015-09-03 15:56:52 Aw, not fast enough 2015-09-03 15:57:14 res_randomid is quite a silly implementation 2015-09-03 15:57:26 I think I could do better and I know nothing about generating random numbers 2015-09-03 15:58:01 ac4random was mentioned on that same bug I believe 2015-09-03 16:00:59 if I want to contribute to aports do I need to have a github account? 2015-09-03 16:01:13 not at all 2015-09-03 16:01:58 have a look at http://wiki.alpinelinux.org/wiki/Creating_an_Alpine_package 2015-09-03 16:02:46 you can send a patch to the aports mailing list 2015-09-03 16:02:56 got it, thanks! 2015-09-03 16:03:22 I'm gonna fix up libasr to get it working with newer automake 2015-09-03 16:04:56 icicled: you want opensmtpd that badly? :) 2015-09-03 16:05:05 yes! 2015-09-03 16:05:31 it's small & simple 2015-09-03 16:05:39 and I need to stand up an email server soonish 2015-09-03 16:08:04 Could you send your patches of libasr upstream to __gilles? (#opensmtpd) that way we don't have to patch libasr at package build time :) 2015-09-03 16:08:36 can do 2015-09-03 16:09:48 icicled: have you seen https://github.com/OpenSMTPD/libasr/issues/8 ? 2015-09-03 16:10:11 yes, I'm starting from there 2015-09-03 16:10:11 =] 2015-09-03 16:11:17 hum, iirc the pids are already quite randomized 2015-09-03 16:12:17 no, that's in *bsd land 2015-09-03 17:20:44 have anyone installed AL on Hyper-V? If yes, how was its behaviour? 2015-09-03 19:31:07 Aw, Alpine is also missing zram 2015-09-03 19:31:16 And rpmextract 2015-09-03 19:32:32 hi 2015-09-03 19:35:53 Heyas TsT 2015-09-03 19:45:01 Ugh, this VM is insanely slow 2015-09-03 19:45:15 Weechat's lag counter is even showing 2015-09-03 19:45:28 Host CPU? Completely idle.. 2015-09-03 19:45:42 <_ikke_> IO? 2015-09-03 19:47:25 Huh, something is writing 2MB/s to disk 2015-09-03 19:47:48 But really, if that amount of IO is going to make my machine behave like this, I may as well stop using it.. 2015-09-03 19:48:35 <_ikke_> Is it swapping? 2015-09-03 19:48:49 Of course, Alpine doesn't have iotop so I have no idea what is writing.. 2015-09-03 19:48:57 _ikke_: Nope 2015-09-03 19:49:06 At least, the host isn't.. 2015-09-03 19:49:15 <_ikke_> It's in testing 2015-09-03 19:49:23 <_ikke_> iotop 2015-09-03 19:50:39 The answer is yes.. my Alpine VM was swapping 2015-09-03 19:50:57 <_ikke_> :-) 2015-09-03 19:50:59 Why does IO suck so much now? It didn't used to when I first started using Arch 2015-09-03 19:51:32 but like.. since a couple of years ago, anything touching the disk just gives me a first class ticket to letsnotrespond-ville 2015-09-03 19:54:39 <_ikke_> Well, swapping is bound to cause slowness, but in general it shouldn't be a problem 2015-09-03 19:55:25 turn off swap :) 2015-09-03 19:58:13 It is. honestly, my entire machine just becomes really laggy 2015-09-03 19:58:43 dalias: Yeah, and that doesn't stop other disk IO stuff from slowing my machine down 2015-09-03 19:59:06 I'd rather fix whatever the problem is, than just cutting off one possible trigger 2015-09-03 19:59:43 But clearly, when one VM wants disk, it is hogging it from all other VMs 2015-09-03 20:00:15 That's clearly not good behaviour :) 2015-09-03 20:00:50 For example, that swapping was writing 2MB/s to disk and made everything unusable 2015-09-03 20:01:20 So if I, say, start playing a 1080p movie in my Linux VM and that is reading 4MB/s or wwhatever.. all my other VMs are going to slow down? 2015-09-03 20:02:36 <_ikke_> What kind of storage do you have? 2015-09-03 20:05:04 SSDs make a huge difference 2015-09-03 20:05:10 seek times 10x better. 2015-09-03 20:26:35 Seek times 100x better, storage space 100x worse 2015-09-03 20:27:07 I have like. 6 HDDs of various sizes between 1TB and 2TB and they are all riding the 97%+ used :P 2015-09-03 20:27:10 And I have umm 2015-09-03 20:27:33 2 * 1TB Western Digital Blue's in RAID1 2015-09-03 20:30:39 Same thing happens with my 2TB Seagate drives that Arch is on 2015-09-03 20:38:04 lol 2015-09-03 21:15:26 is busybox mount giving "failed: No such device" when trying to mount a NFS share a known issue ? 2015-09-03 21:15:37 using edge if that matters ... 2015-09-03 21:17:04 mount: unknown filesystem type 'nfs' 2015-09-03 21:17:17 -_- 2015-09-03 21:17:38 what command are you using? 2015-09-03 21:17:52 mount -t nfs IP:/path /mntpoint ? 2015-09-03 21:18:19 busybox looks like it supports nfs mounts 2015-09-03 21:18:48 might be a kernel thing though 2015-09-03 21:19:16 coredumb: ^ does that help at all? 2015-09-03 21:19:40 seems like I'm still running an old kernel -_- 2015-09-03 21:24:40 nfs support is in the kernel, and apparently kernel had to have been copmiled with nfs_v3 support if you need it 2015-09-03 21:24:44 otherwise v2 might work? 2015-09-03 21:24:58 oh it's telling you it doesn't even know the filesystem 2015-09-03 21:25:01 odd 2015-09-03 21:31:14 I had a similiar issue on voidlinux... 2015-09-03 21:31:46 maybe this helps: https://github.com/voidlinux/void-packages/issues/1521 2015-09-03 21:33:00 that looks like it's targeting nfs vers 4.2 2015-09-03 21:33:44 I'd think he could sepcify -o vers=3 2015-09-03 21:37:31 Sleaker: yep was because of my upgraded kernel not rebooted 2015-09-03 21:37:33 silly me -_- 2015-09-03 21:37:40 ah 2015-09-03 21:37:43 got it working? 2015-09-03 21:51:44 Sleaker: yes all good 2015-09-03 21:54:40 just rebooting on 3.18.20 made my interface get funky names wlp3s0 2015-09-03 21:54:43 O_o 2015-09-03 21:54:48 It makes me feel better knowing other people make the same sort of silly mistakes I do :P 2015-09-03 21:55:05 They are persistent names 2015-09-03 21:55:20 yeah but where are they coming from 2015-09-03 21:55:25 Umm 2015-09-03 21:55:28 surely not just from the kernel upgrade 2015-09-03 21:55:45 Could be, I don't remember where it got introduced 2015-09-03 21:56:27 got other machines with the same kernel with the persistent names :/ 2015-09-03 21:56:37 I get that issue when I boot Arch. I get eth0 and eth1 and they swap because parallel boot so important 2015-09-03 21:56:44 I hate those persistent names 2015-09-03 21:56:47 So I have to bridge them both 2015-09-03 21:57:35 They never swap in Alpine :) 2015-09-03 21:58:03 Or they haven't in the 1000 times I've booted it in the last couple of weeks! 2015-09-03 21:58:34 grumpf 2015-09-03 22:03:31 CandyAngel: probably udev update I guess 2015-09-03 22:08:39 eudev* 2015-09-03 23:22:53 ACTION headdesks 2015-09-03 23:22:57 Another roadblock 2015-09-03 23:23:34 Both my graphics cards are in the same IOMMU group -.- 2015-09-03 23:41:35 I can't even try moving it into another slot because it won't fit in the other slot it can go in >.< 2015-09-04 01:51:50 mh, why is alpine generating a DSA openssh key on setup :/ 2015-09-04 02:47:46 CandyAngel: is that a motherboard thing? 2015-09-04 03:44:04 Query: I've seen the wiki regarding Alpine Linux as a DMVPN spoke node, I'd like to ask if an alpine box can be used as a hub node as well, and how you go about trying to achieve that. 2015-09-04 04:30:02 I'm installing Alpine to a virtualbox vm, having a rough time seems to get stuck at both the repo updating and the copying to the drive. Is there a minumum system requirements for virtualbox on this one? 2015-09-04 04:41:41 nevermind, it's stopping in random places now... I'll sleep on it, this is retarded lol, night all. 2015-09-04 06:43:34 I would need a bit of help in getting a new initramfs for armhf. I am trying to do the not supported thing of getting / on ext4. I have modified setup-disk and got it to copy everything to my new root partition. Now I would need to modify the initramfs.rpi2 to use the new root device 2015-09-04 06:44:04 I know it is not supported, but I would definitely want to run off disk rather than off ramdisk. 2015-09-04 06:45:11 My layout is like this: 200MB VFAT (for booting), 2G Swap, 12,7G EXT4 / 2015-09-04 06:45:42 The root partition is ready populated 2015-09-04 06:45:49 I am able to chroot into it 2015-09-04 06:49:19 engblom: I've just unplugged my RPi yesterday as i needed a spare SD card - :( 2015-09-04 06:50:11 :( 2015-09-04 06:50:34 you shouldn't need to modify setup-disk, you can just pass it a mount point 2015-09-04 06:51:14 ScrumpyJack: setup-disk does not accept /boot as vfat. Also it tries to write a MBR and it tries to install syslinux 2015-09-04 06:51:58 ah yes, of course 2015-09-04 06:52:18 The initramfs it created does not work, so there something manual has to be done 2015-09-04 06:53:08 Sadly, my knowledge of initramfs is very lacking 2015-09-04 06:53:35 perhaps you could unpack the initramfs and look at what it does. you should be able to modify it, repack it, and replace it on the SD card, as it's not a iso or anything annoying. 2015-09-04 06:54:28 it's a gzipped cpio 2015-09-04 06:54:43 Ok, I might give that a try 2015-09-04 06:54:49 right. coffee 2015-09-04 07:07:18 Sleaker: Yeah.. in both cases 2015-09-04 08:00:35 <^7heo> Moin @ 2015-09-04 08:01:28 <^7heo> On my system (setup with 3.2.3, using setup-alpine and then setup-xorg-base, xinit freezes the system unless I manually start udev 2015-09-04 08:01:46 <^7heo> udev is present in the rc-update output with a sysinit trigger. 2015-09-04 08:02:05 <^7heo> Is that a bug? A misconfiguration? 2015-09-04 08:04:24 dunno 2015-09-04 08:05:07 it needs udev to get hotplugging of keyboard and mouse 2015-09-04 08:05:27 i think you can use xorg without udev but then you need to add xorg.conf for keyboard and mouse 2015-09-04 08:11:03 ^7heoooooooooo 2015-09-04 08:11:06 ACTION waves 2015-09-04 08:15:35 <^7heo> hey CandyAngel 2015-09-04 08:16:04 <^7heo> ncopa: sure; I was just wondering why isn't udev started while it is in rc-update. 2015-09-04 08:16:42 ^7heo: what runlevel? 2015-09-04 08:16:49 <^7heo> sysinit 2015-09-04 08:16:56 <^7heo> that's what I meant by "trigger" 2015-09-04 08:17:13 <^7heo> I wasn't aware that they were called runlevels such as with sysvinit 2015-09-04 08:17:51 <^7heo> ncopa: should you need any logfile, please feel free to ask me for a paste. 2015-09-04 08:17:52 "as with unix" :)) 2015-09-04 08:18:19 <^7heo> darkfader: I dunno, I feel it's different; but I haven't given it much toughts/ 2015-09-04 08:18:24 <^7heo> s/\/$/./ 2015-09-04 08:20:03 and udev is not started automatically at boot? 2015-09-04 08:20:23 <^7heo> for all I know, no. 2015-09-04 08:20:45 <^7heo> all I can say is that when I /etc/init.d/udev status I get *stopped* 2015-09-04 08:21:04 <^7heo> (after `login`, ofc) 2015-09-04 08:22:02 <^7heo> and the only udev line I have in `dmesg` is the one corresponding to when I started it. 2015-09-04 08:22:16 <^7heo> should I try changing the runlevel of udev? 2015-09-04 08:22:47 no, sysinit should be correct 2015-09-04 08:22:56 <^7heo> I mean, for testing purposes. 2015-09-04 08:23:20 it coudl be openrc cache and timestamps issue 2015-09-04 08:23:47 there is a command to force regenerate the openrc dep cache 2015-09-04 08:23:52 <^7heo> also; while I'm at it, the default configuration for wpa_supplicant is bogus: there's a `freq_list=5180` not commented in the middle of the file; limiting the APs that wpa picks when scanning. 2015-09-04 08:23:53 i dont remember what it was 2015-09-04 08:24:00 <^7heo> should I open a bug for the last one? 2015-09-04 08:24:12 if default config is bogus, then yes 2015-09-04 08:24:16 <^7heo> okay. 2015-09-04 08:24:22 <^7heo> gonna do that now. 2015-09-04 08:31:22 <^7heo> ncopa: #4612 created. 2015-09-04 09:25:47 ^7heo: Yesterday I noticed one thing: setup-xorg-base can not be run over ssh. I am new to alpine myself, so I am no expert here, but I wanted to point out the mistake I did myself, in case it would be the same you did. 2015-09-04 09:30:22 ACTION noticed another missing package preventing him from switching his main workstation to alpine: blender 2015-09-04 09:38:21 <^7heo> engblom: nah I'm not running that over ssh. 2015-09-04 09:38:57 <^7heo> and I don't get what cannot be run over ssh, btw. 2015-09-04 09:39:04 <^7heo> (I just read the script) 2015-09-04 09:39:25 <^7heo> maybe the setup restarted your network, but other than that I don't see what can cause problems. 2015-09-04 09:47:31 ^7heo: It is the network restarts killing it 2015-09-04 10:53:07 good news everyone - I've got opensmtpd compiling without any errors 2015-09-04 10:53:38 it's a hack job but it *appears* to work 2015-09-04 10:54:27 Nicely done joncfoo :) 2015-09-04 10:55:21 let me put up the dirty, dirty patch online 2015-09-04 10:56:21 Is it better to have dirtypatched and compiled than to never have compiled at all? 2015-09-04 10:58:12 I reckon 'tis best to have dirtypatched at least once 2015-09-04 10:59:30 <_xs> ^7heo: i had the same problem with udev's service, its effectively working removing the cache and generating a new one 2015-09-04 10:59:33 joncfoo: cool. icicled was trying to do the same lat night too 2015-09-04 11:01:16 yep that's me :) 2015-09-04 11:01:23 https://gist.github.com/joncfoo/2a9c0ab0d9cd899d6791 2015-09-04 11:01:24 ay! 2015-09-04 11:03:15 I put a comment at the bottom as to what needs to be done 2015-09-04 11:06:57 that's neat 2015-09-04 11:07:01 I know just enough C to throw stuff together & make things work - any pointers in improving the patch would be appreciated (so punny) 2015-09-04 11:08:18 that's beyond my skillset. have you got a APKBUILD file for opensmtpd? 2015-09-04 11:08:40 ScrumpyJack: no, not yet - I'll get to it this weekend 2015-09-04 11:09:46 I spent last night and this morning working through the errors and figuring out how autoreconf works 2015-09-04 11:10:26 joncfoo: great work! 2015-09-04 11:10:42 I need to test if it actually works :P 2015-09-04 11:10:46 did it compile with libasr from the AL repos? 2015-09-04 11:11:13 I'm thinking about donating some money to alpine, even if I'm not even an alpine user ;) 2015-09-04 11:11:25 no, I had to make a couple changes to libasr 2015-09-04 11:11:36 to include res_randomid 2015-09-04 11:11:57 Wait.. Alpine accepts donations? 2015-09-04 11:12:05 I hope so :) 2015-09-04 11:12:08 As our voidlinux port to musl-libc is largely based on alpines patches. 2015-09-04 11:12:43 qq, is voidlinux ports based + musl ? 2015-09-04 11:13:18 voidlinux builds against musl and glibc. nutters 2015-09-04 11:13:25 :) 2015-09-04 11:13:47 soon musl by default for voidlinux? 2015-09-04 11:13:49 ;) 2015-09-04 11:14:41 I don't think so. :) 2015-09-04 11:16:10 we should put together some t-shirts and stickers 2015-09-04 11:18:17 how is al organized? 2015-09-04 11:19:05 http://bugs.alpinelinux.org/projects/alpine 2015-09-04 11:19:20 Is there some kind of foundation in the background? 2015-09-04 11:19:22 Gottox: ^^ 2015-09-04 11:21:10 thx. 2015-09-04 11:27:29 right, have a good weekend folks 2015-09-04 11:31:25 you too 2015-09-04 11:33:17 Baibai ScrumpyJack 2015-09-04 11:37:25 Apparently stable releases of grsec to the public are going to cease in 2 weeks. 2015-09-04 11:37:40 https://grsecurity.net/announce.php 2015-09-04 11:52:00 Huh 2015-09-04 11:52:30 The part apart a company using an old test patch is.. weird 2015-09-04 11:52:47 I can understand why they would be upset by it, but the company is using something grsecurity put their name to 2015-09-04 11:53:00 If it doesn't meet their standards, they need to separate it 2015-09-04 11:53:12 Test stuff is grsecurity 2015-09-04 11:53:23 Stuff that meets their standard is grsecurity *certified* 2015-09-04 11:55:11 Then if that company said it was certified when it wasn't, then they would have grounds to sue under misrepresentation or whatever :\ 2015-09-04 11:55:49 Also 2015-09-04 11:55:58 What they are doing wouldn't even solve that problem 2015-09-04 11:56:15 That company would still be able to grab "unfit" public code and use it and say it was grsecurity :\ 2015-09-04 11:58:58 Maybe I'm misunderstanding? But it seems like they've destroyed a bridge in an attempt to stop an aerial attack 2015-09-04 12:17:13 <^7heo> damn 2015-09-04 12:17:25 <^7heo> that grsecurity notice is very saddenning... 2015-09-04 12:26:17 what are the implications for alpine, if any? 2015-09-04 12:27:54 <^7heo> joncfoo: I guess, the necessity to provide only latest kernels. 2015-09-04 12:28:38 <^7heo> I am not working in close to the team providing the grsec kernel in alpine, but my assumption would be that it might cost us stability. 2015-09-04 12:28:56 <^7heo> s/in/any/ 2015-09-04 12:29:18 <^7heo> thanks, intel. 2015-09-04 13:05:34 just asked libasr to be updated upstream - here goes nothing 2015-09-04 13:08:03 ACTION wonders how difficult it would be to get Blender ported for Alpine. As ncopa said he will probably work on Thunderbird, Blender is probably the last stopper for me before I can switch my work station over to Alpine. 2015-09-04 13:10:26 yay, thunderbird o/ 2015-09-04 13:22:38 <^7heo> engblom: are you familiar with abuild? 2015-09-04 13:22:50 ^7heo: Not yet 2015-09-04 13:22:56 <^7heo> get familiar with it ;) 2015-09-04 13:23:00 <^7heo> it's incredibly easy to use. 2015-09-04 13:23:21 At the moment I have downloaded blender, trying to see what dependencied are already existing as apk, and then I will try to build 2015-09-04 13:23:26 <^7heo> dont' start wtih blender. 2015-09-04 13:23:46 <^7heo> start with something smaller. 2015-09-04 13:23:54 <^7heo> support a very small software that we don't have yet. 2015-09-04 13:24:00 <^7heo> it's not hard to find :) 2015-09-04 13:24:17 <^7heo> go for terraform, for example. 2015-09-04 13:24:28 <^7heo> it's VERY easy to build, so that should get you started. 2015-09-04 13:24:33 But Blender is the software I need at the moment. The rest I can somehow manage without, even though they would be nice to have 2015-09-04 13:24:52 <^7heo> yeah, look: alpine linux is not a whine'n'get distribution. 2015-09-04 13:24:59 <^7heo> so either you wait silently or you get busy. 2015-09-04 13:25:15 <^7heo> and honestly, I'd rather get busy if I were you. 2015-09-04 13:25:17 That is why I began trying to see if I get blender to work myself 2015-09-04 13:25:26 <^7heo> sure, that's the good attitude. 2015-09-04 13:26:05 <^7heo> but, honestly, I think it's a bit too steep of a step, starting fresh... 2015-09-04 13:26:18 <^7heo> that's the exact reason I am telling you to start with something smaller. 2015-09-04 13:26:41 <^7heo> I suggested `terraform`, which I already compiled myself for one of my previous workstations; I know it to be very easy to do. 2015-09-04 13:26:58 <^7heo> writing an abuild for it should get you started and shouldn't take long. 2015-09-04 13:27:38 <^7heo> in the end if you do that it will save you time, because you will know how abuild work with a simple thing, and you will be able to focus your efforts in getting blender to build, not getting blender to build AND understanding abuild at the same time. 2015-09-04 13:28:04 <^7heo> If you want I might be able (please note the 'might') to assist you the coming week end. 2015-09-04 13:28:12 <^7heo> (i.e. tomorrow and sunday) 2015-09-04 13:28:52 ^7heo: I understand what you mean, but at this moment I rather trying to compile it directly to see if I at all get it to compile without making huge changes. If I get it to compile, then I will look at abuild (either for blender or a smaller tool). If I do not get Blender to compile, I do not need to spend time learing how abuilds work and I have quite little extra time. 2015-09-04 13:30:14 I have a couple of suggestions for some packages o/ 2015-09-04 13:30:36 <^7heo> TheLemonMan: yeah, if engblom wants, you might even get that done while he learns :) 2015-09-04 13:30:51 <^7heo> engblom: understood. Learn from your own mistakes. ;) 2015-09-04 13:56:47 Seem to be a strange problem: cc is running out of memory while compiling some of the files for Blender 2015-09-04 13:57:40 Do alpine have a restriction, in similar way to bsd, for how much memory a process is allowed to use? 2015-09-04 13:59:22 So far I have only seen memory restrictions in BSD 2015-09-04 14:00:27 engblom: cgroups can be used for it, it is extremly awkward to configure but does work, including cpu restrictions 2015-09-04 14:01:29 darkfader: If I want to give one user unlimited of resources, how would I quickly do it? 2015-09-04 14:02:37 idk, i didnt ever need to change it 2015-09-04 14:03:03 i tested memory-restricting my shell once using cgroups but that was 2 years ago 2015-09-04 14:04:35 I got 81% of blender to compile... so it did not work without some extra work. I will have to continue another time if I get time 2015-09-04 14:05:22 81% of blender is quite a lot :) 2015-09-04 14:05:31 blender-lite 2015-09-04 14:05:51 Not the full one with physics and movie editing. 2015-09-04 14:26:35 Is there any way for me to find out if the stock Alpine kernel has the ACS override patch applied? 2015-09-04 14:30:31 CandyAngel http://hastebin.com/ubarunumit.avrasm 2015-09-04 15:04:40 BitL0G1c: Thankies 2015-09-04 18:00:20 ncopa: i installed alpine non hardened on a vps. They dont have the hardened. Is it possible to utdate? 2015-09-04 18:00:51 Actually not a vps. But arm physicall 2015-09-04 18:08:12 https://www.irccloud.com/pastebin/XWmh5nLy/ 2015-09-04 18:08:17 Anyone know how to resolve this one? hehe 2015-09-04 18:18:44 How can you choose which kernel to boot from? :O 2015-09-04 18:20:57 Edit the bootloader's configuration. 2015-09-04 18:21:15 It's probably /boot/extlinux.conf or syslinux.cfg or something like that. 2015-09-04 18:21:22 Then change the 'LINUX' line. 2015-09-04 18:21:34 You may need to change the initramfs as well. 2015-09-04 18:23:26 shardz: my /boot only consists of 2015-09-04 18:23:29 https://www.irccloud.com/pastebin/0YekeUq9/ 2015-09-04 18:24:42 <_ikke_> config-grsec I believe 2015-09-04 18:25:15 _ikke_: thats just the build conf from the kernel 2015-09-04 18:40:50 any takers? :P :D 2015-09-04 18:45:02 It doesnt look like i have extlinux 2015-09-04 20:45:40 anyone here familiar with autoconf/autotools? 2015-09-04 21:29:34 anyone know what package provides user utilities such as usermod? 2015-09-04 21:30:03 I'm going to have to put my project on hold.. apparently, neither of my GPUs have EFI ROMs, so they don't really work for VGA passthrough 2015-09-04 21:30:19 Huuuuuuuuge thank you to you all for helping me where you could :) I really appreciate it 2015-09-04 22:09:27 hm is there an armhf docker image or do I need to build it myself 2015-09-04 23:36:14 ammunta: a quick Google search shows there are at least a couple: https://hub.docker.com/r/armbuild/alpine/ https://hub.docker.com/r/armhfbuild/alpine/ 2015-09-04 23:41:38 andyshinn: so, yes, but I still need to build it myself basically 2015-09-04 23:48:56 Anyone knows how to change kernel? No extlinux pkg availavle 2015-09-05 10:59:40 hello 2015-09-05 11:00:01 ls 2015-09-05 11:00:05 ups 2015-09-05 11:01:17 hey sorry but i'm new at alpine and i'm tryng to install nginx i have done all (i think) but when i try to curl http://localhost or curl http://127.0.0.1 2015-09-05 11:01:26 appears connection refused 2015-09-05 11:01:47 how can i solve that please 2015-09-05 11:01:59 thanks a lot 2015-09-05 11:02:55 <_xs> is the service started ? 2015-09-05 11:03:56 dono sorry is the first steps 2015-09-05 11:04:00 what service nginx? 2015-09-05 11:04:35 yapo 2015-09-05 11:04:36 yes 2015-09-05 11:04:39 bash-4.3# service nginx start 2015-09-05 11:04:39 * WARNING: nginx is already starting 2015-09-05 11:05:18 curl http://127.0.0.1:80 2015-09-05 11:05:19 curl: (7) Failed to connect to 127.0.0.1 port 80: Connection refused 2015-09-05 11:05:31 <_xs> any firewall ? 2015-09-05 11:06:16 i think wall is instaled 2015-09-05 11:06:46 <_xs> maybe you have loaded a default ruleset 2015-09-05 11:07:21 <_xs> awall list -a 2015-09-05 11:08:03 command not found :| 2015-09-05 11:08:11 <_xs> then awall isn't installed 2015-09-05 11:08:16 yap 2015-09-05 11:08:25 why failed connect? 2015-09-05 11:08:31 <_xs> try this, grep listen /etc/nginx/nginx.conf 2015-09-05 11:08:34 if i don't have awall 2015-09-05 11:08:58 <_xs> if it listen to 80, check iptables -vL 2015-09-05 11:09:11 http://pastebin.com/ACzEPzfv 2015-09-05 11:12:19 can't initialize iptables filter 2015-09-05 11:13:18 sorry _xs 2015-09-05 11:13:47 <_xs> which kernel are you using ? 2015-09-05 11:14:11 <_xs> you can check in /var/log/messages and service nginx status 2015-09-05 11:14:57 Linux e5e8134042da 3.18.11-hypriotos-v7+ #2 SMP PREEMPT Sun Apr 12 16:34:20 UTC 2015 armv7l Linux 2015-09-05 11:15:46 <_xs> seems no netfilter enabled ;) 2015-09-05 11:15:56 i'm runnig docker container 2015-09-05 11:16:02 <_xs> ow.. 2015-09-05 11:16:28 <_xs> this thing 2015-09-05 11:16:51 :D 2015-09-05 11:16:52 lolll 2015-09-05 11:18:14 <_xs> sorry guy, i'm pretty incompetant in that kind of network virtualization 2015-09-05 11:18:46 <_xs> but it seems that the host's kernel drop your packets 2015-09-05 11:19:35 <_xs> maybe because the port used elsewere or anything 2015-09-05 11:20:35 <_xs> ACTION was thinking cgroups able to isolate network stacks 2015-09-05 11:21:36 <_xs> fcp14: have you tried running nginx behing another port ? 2015-09-05 11:21:41 <_xs> behind* 2015-09-05 11:22:47 gona check 2015-09-05 13:05:02 <_xs> ldd: /usr/bin/firefox: Not a valid dynamic program 2015-09-05 13:05:06 <_xs> makes me laugh 2015-09-05 13:06:18 <_xs> oh it's a script ! 2015-09-05 21:07:22 i partitioned formated and mounted my filesystems the way i wanted. but it appears i cant just mount the new rootfs and install alpine 2015-09-05 21:09:17 i remember once i added a hdd did a setup-disk there and then copied the OS to my own FS 2015-09-05 21:09:38 but that isnt very elegant... 2015-09-05 22:24:45 setup-disk /mnt denies to work if /mnt is xfs?! 2015-09-06 00:00:33 I'm running 3.2.3 Alpine desktop and when I launch firefox, there's no text. Anyone have an idea what the issue might be? Its hard to use the browser when the UI has no text. Webpages have text, just the UI shows icons only. 2015-09-06 00:05:33 midori has text in its UI 2015-09-06 01:34:33 ncopa: What's the strategic plan for dealing with the upcoming lack of grsec stable releases? 2015-09-06 14:31:51 I'm creating a package for opensmtpd and the default users for it are named _smtpd & _smtpq - is that ok? 2015-09-06 14:31:57 or should I remove the prefixes? 2015-09-06 14:32:28 I think openbsd conventions are that an underscore denotes a system/daemon user 2015-09-06 15:10:12 joncfoo, we have not used underscore for user/group names in alpine 2015-09-06 15:17:42 fabled: Thanks, I saw there is an openjdk package for armhf now. I do have my rpi2 at work so I will not get to test until tomorrow. 2015-09-06 15:18:24 engblom, yes, it got built sometime yesterday or so 2015-09-06 15:22:46 fabled: You said there is some way to run armhf port directly off SD. Officially it seem to not be possible as setup-disk does not want to install to SD-card. I modified a bit setup-disk to accept vfat boot partition, no syslinux and no MBR. I got stucked with getting a working initramfs. Have you been running yourself / from SD? 2015-09-06 15:23:24 engblom, you'd need to partition the SD; have first partition FAT / boot; and second partition as ext4 or similar 2015-09-06 15:23:47 fabled: That I did. I have a working root partition on ext4 and I can chroot into it 2015-09-06 15:23:56 I just have problem to get it to boot 2015-09-06 15:24:25 My layout is 200M vfat, 2Gb swap, 12.7Gb ext4 2015-09-06 15:26:46 the rpi /boot is special. you should just keep using kernel+initramfs+other files from the dist .tar.gz 2015-09-06 15:27:04 we should fix setup-disk / update-kernel to do the right thing in arm 2015-09-06 15:28:22 One alternative would be to provide a disk image. One could always change the size of the partitions after writing the image to SD card with gparted or similar tool. 2015-09-06 15:30:33 we prefer to ship the tmpfs / lbu based default images 2015-09-06 15:31:20 but understandably it does not work with all installs on arm / when you need several gigs of stuff in root 2015-09-06 15:36:35 Yes, and this will be the case for me. And also for a friend of mine with rpi2 2015-09-06 15:38:02 For the moment I manage with raspbian, but coming from BSD world, I definitely like alpine better than the "over-gnuified" distros. 2015-09-06 16:25:04 fabled: ok, I'll fix the default users it expects. thanks! 2015-09-06 22:33:09 ACTION eats Adran 2015-09-07 02:14:53 So I figured out how to upgrade to the edge packages... looks like as long as I disable the wfb module for Xorg xf86-video-vesa loads now... firefox looks all screwed up though for whatever reason... 2015-09-07 02:16:51 it doesn't seem to be a gtk problem though... asl pidgin works :) 2015-09-07 02:19:51 Its very heartening to be be able to run pidgin in 34MB ram... It appears that on my gentoo system pidgin has nearly twice that resient in memory O.o 2015-09-07 02:34:04 It would be nice if there were a netsurf package... but I guess dillo will do in the meantime. Firefox is a bit bloaty on a IBM 240x (I intend to do the 256Mb ram hack on it) an put alpine linux on there. 2015-09-07 07:22:50 Hi! Can someone verify if vpnc works with alpine? 2015-09-07 07:24:01 We had a bugreport at voidlinux and I suspect, that alpine is affected too. 2015-09-07 07:24:37 see https://github.com/voidlinux/void-packages/pull/2472 2015-09-07 07:53:09 morning 2015-09-07 08:19:30 https://github.com/scaleway/kernel-tools/issues/142 2015-09-07 08:19:33 :-) 2015-09-07 08:19:43 Im not affiliated with them but looks good 2015-09-07 08:33:01 i've got a miniserver there, atm it's running debian for sw build purposes but i have run it with alpine first 2015-09-07 08:33:08 worked like a charm 2015-09-07 08:33:21 btw they use nbd for disk storage 2015-09-07 08:33:50 darkfader: ah, so not actual local storage 2015-09-07 08:47:15 what does this mean for Alpine Linux? https://grsecurity.net/announce.php 2015-09-07 08:52:57 ScrumpyJack: maybe we should just set up something with him that officially covers alpine... but what if someone does the same mess with alpine's kernel? 2015-09-07 09:00:07 ScrumpyJack: no official statement yet, but ncope said he would email spender/pax and ask 2015-09-07 09:14:58 darkfader: the only thing i miss is custom kernels on scaleway, after that I move my servers 2015-09-07 09:15:04 darkfader: hopefully they dont go bankrupt tho :p 2015-09-07 09:20:04 hmm, since i upgraded my hardware (desktop at work), my NIC is up and down like a yoyo 2015-09-07 09:20:28 Intel 82579LM card with e1000e driver 2015-09-07 09:21:52 e1000e version: 2.3.2-k in edge 2015-09-07 10:15:17 where are the scaleaway servers? it's not mentioned on their site 2015-09-07 12:51:39 I would want to try the new edge version of armhf. How should I do? The problem is that 'apk update; apk upgrade' takes a lot of memory. 2015-09-07 12:51:51 I would want to test openjdk for armhf 2015-09-07 12:52:02 It is only existing for edge. 2015-09-07 12:54:01 you could mount var to sd 2015-09-07 12:54:25 if you are on rpi that is 2015-09-07 12:54:38 Yes, rpi it is. 2015-09-07 12:55:55 Mounting /var would keep the apk cache, but how about keeping all the upgraded packages extracted to / ? 2015-09-07 12:58:11 running openjdk on tmpfs on an rpi is usually problematic 2015-09-07 12:58:29 not sure how big that apk is 2015-09-07 12:59:24 That means I will still have to wait with alpine for my rpi... The whole purpose of my rpi2 is to run Clojure programs controlling stuff. 2015-09-07 12:59:32 not sure how apk handles an upgrade. i think it creates tmp files and when ready replaces the original files. 2015-09-07 13:38:13 engblom: does apk update && apk upgrade fail on your RPi? 2015-09-07 13:39:33 ScrumpyJack: I am testing at the moment. However, openjdk got already something like 90 dependencies and it is big itself. Leiningen (the build tool for Clojure) require librssl/openssl and that adds quite a bit of dependencies 2015-09-07 13:40:04 Adding all that to tmpfs takes quite hard on ram.... then also some ram is needed for actually running stuff 2015-09-07 13:42:23 can we see a df of your RPi? 2015-09-07 13:49:19 should be the half of available memory 2015-09-07 13:49:28 tmpfs uses half 2015-09-07 13:49:43 so it should be close to 500 on an rpi 2015-09-07 13:50:07 atleast its the default. 2015-09-07 14:32:40 git commits info on frontpage is frozen since 20/8 2015-09-07 14:33:04 I'm not sure if it was reported already 2015-09-07 15:01:25 I got openjdk7 installed by creating a chroot environment on the SD-card. 2015-09-07 15:02:29 There is a problem with openjdk7 for armhf and the keystore. It is empty. 2015-09-07 15:33:43 engblom: you know how do add a cert to it 2015-09-07 15:33:45 ? 2015-09-07 15:33:56 can help you later tonight if you struggle 2015-09-07 16:15:56 engblom: I can't install a DE + openjdk on the default tempfs cauese it's too small 2015-09-07 16:16:06 openjdk tries to pull down like 250MB+ of stuff 2015-09-07 16:16:51 I was going to look into overlay/squashfs type deal for using with USB boot 2015-09-07 16:28:03 darkfader: No, I do not know. I am at home, and my rpi2 is at work so I will not have a chance to test again until tomorrow. 2015-09-07 16:30:11 Sleaker: The best would be to have setup-disk properly updated to allow installing to SD-card. When it comes to big things like java stuff on small computers like rpi I think running / from SD would be the only right. 2015-09-07 16:30:43 Sleaker: I slightly modified setup-disk to get it to install my chroot without trying to install syslinux and MBR. 2015-09-07 16:31:02 engblom: possibly. 2015-09-07 16:31:13 I still struggle with getting it to boot straight into my new system 2015-09-07 16:31:36 the issue is that in my use case I have written the ISO to USB, and setup the system for lbu to commit changes to that. 2015-09-07 16:31:57 and you can't use that as a standard partition to install to. 2015-09-07 16:32:08 so I'm not sure setup-disk would even work 2015-09-07 16:32:42 would need to have a full new setup or a way to re-master the image to have a bunch of packages installed on the base USB to begin with 2015-09-07 16:33:04 but I think alpine mounts the tempfs as / 2015-09-07 16:33:10 so that presents a slight problem 2015-09-07 16:33:44 I hope someone with distro-maintaining skills will take upp setup-disk for armhf as a project. While I am having scripting skills to edit the script, I lack the knowledge to get a rpi system to boot. 2015-09-07 16:33:59 although, I could format the USB drive as ext4 right? 2015-09-07 16:34:25 then I could just install to the USB as a normal install. 2015-09-07 16:34:43 Sleaker: That should be possible, at least if you talk about amd64 2015-09-07 16:37:17 yah x86/amd64 2015-09-07 16:38:16 I don't have an arm device to paly with yet 2015-09-07 16:38:30 might pick one up in the future when the platform has a bit better GL support. 2015-09-07 16:39:16 I take that back, I have a Nexus 7 2012 with Cyanogen mod on it and an Ouya. 2015-09-07 16:39:44 although I don't think the Ouya can really have anything done to it cause they didn't include a proper recovery mode in the firmware. 2015-09-07 16:41:15 Personally I do not care about GL support for a limited computer like rpi. I bought the rpi because of the GPIO port 2015-09-07 16:41:22 right 2015-09-07 19:05:38 hello 2015-09-07 19:06:25 hi ;) 2015-09-07 19:06:54 I have an instance of alpine linux running on a scaleway server, just set it up with setup-alpine & found I cant login after reboot 2015-09-07 19:07:21 I switched to using OpenSSH during the setup process 2015-09-07 19:07:35 when I ssh in, I see: PTY allocation request failed on channel 0 2015-09-07 19:11:27 venture37: maybe it's missing /dev/pts? 2015-09-07 19:12:03 indeed 2015-09-07 19:12:23 Is this right for adding to fstab? 2015-09-07 19:12:24 devpts /dev/pts devpts gid=5,mode=620 0 0 2015-09-07 19:14:53 venture37: if you enable the devfs service it should also do it 2015-09-07 19:14:58 without an fstab entry 2015-09-07 19:17:04 service devfs on 2015-09-07 19:17:14 ssh in & same errer 2015-09-07 19:40:07 devpts /dev/pts devpts gid=5,mode=620 0 1 2015-09-07 19:40:27 can't set it to pass 0 as root fs is not there 2015-09-07 19:44:28 I'm talking nonsense 2015-09-07 20:12:34 activated devfs on boot using rc-update add devfs 2015-09-07 20:13:15 commented out the fstab entry 2015-09-07 20:13:20 rebooted 2015-09-07 20:13:36 devpts is now mounted but I get PTY allocation request failed on channel 0 2015-09-07 20:13:37 shell request failed on channel 0 2015-09-07 20:32:35 for someone familiar with ubuntu server is alpine linux very different to learn? 2015-09-07 21:27:18 hydrajump: depends on how you use it. i'm inclined to say "no, should be fine, alpine is more straightforward and cleaner in design" 2015-09-07 21:27:48 if you started with the boot to ram mode you'd say i'm wrong 2015-09-07 21:27:56 but normal install to disk, should be fine 2015-09-07 21:39:06 resolved my isssue, I'm running a Alpine on a arm64 machine, when you run setup-alpine, it adds acpid to startup, this prevents devpts filesystem from being mounted 2015-09-07 21:39:40 removed the service with rc-update delete acpid & everything works 2015-09-07 22:30:25 darkfader: cool I'm planning on using it installed from disk for a postfix/dovecot server that's currently running on ubuntu 2015-09-07 22:54:17 hydrajump: honestly i think you'll be lot better off. i've spent the day automating ubuntu installs because a friend likes it and i think it'll cost me a few days of sanity :) 2015-09-08 00:34:29 why is the version of tar for alpine-linux not allow --strip-components 2015-09-08 00:34:40 err why does 2015-09-08 00:47:58 atomi: the Alpine version of tar comes from BusyBox, which doesn't support --strip-components 2015-09-08 00:48:09 you can install GNU tar though, using apk add tar 2015-09-08 00:50:33 venture37: I encountered the same issue today. I upgraded mkinitfs from 2.7.1 to the latest git. Reverting http://git.alpinelinux.org/cgit/mkinitfs/commit/?id=ac397ebafd39a277ca94cd37b70ad15d1c1e3d46 got my /dev/pts back. 2015-09-08 00:51:18 There's something in /etc/init.d/devfs that doesn't maintain /dev/pts when remounting /dev. I haven't pinned it down exactly. 2015-09-08 00:52:19 And having /dev as a tmpfs rather than a devtmpfs makes it work for some reason. 2015-09-08 00:53:06 andyshinn, thanks! 2015-09-08 00:53:26 apk add tar should have been my first try 2015-09-08 05:58:50 l 2015-09-08 05:58:50 x 2015-09-08 05:58:54 Oops, sorry. 2015-09-08 06:09:01 icedwater: Hey. Long time no see. 2015-09-08 06:48:08 teiresias: hello :) Can't remember what you helped me with before, but thanks :) 2015-09-08 07:19:00 morning 2015-09-08 07:21:15 i wonder, does anyone run alpine-xen as a desktop? Thinking about using the alpine linux Dom0 as a desktop to be able to have more control over the VMs that I also run on te same desktop with qemu/lxc 2015-09-08 07:23:50 ScrumpyJack: i considered that some time ago but ended up with regular alpine and qemu/libvirt/virt-manager for virtual 2015-09-08 07:27:17 ncopa: do we have cgroups memory control now? 2015-09-08 07:36:21 ScrumpyJack: what was the kernel config knob for that? 2015-09-08 07:37:22 MEMCG: http://tpaste.us/AO16 2015-09-08 07:37:29 yes its enabled for x86_64 2015-09-08 07:40:12 also, i find xen more useful when you want to bind cpus to VMs 2015-09-08 07:40:38 ncopa: what do you think of this? #4617 2015-09-08 07:40:47 I've not had that problem 2015-09-08 07:41:34 not sure 2015-09-08 07:42:09 i have always thoucht that booting up a container should behave similar to booting up real hw (or virtual) as close as possible 2015-09-08 07:42:17 so you need run setup-alpine 2015-09-08 07:42:31 thats how i have thought so far at least 2015-09-08 07:42:48 but i admit that simply enabling networking by default would be handy 2015-09-08 07:44:12 ncopa, what's the best way with pingu to failover to the first ISP once is back on line? 2015-09-08 07:44:24 pinguctl does allow that 2015-09-08 07:44:30 might be an useful feature to add 2015-09-08 07:45:54 fcolista: best way is to let pingu handle that himself 2015-09-08 07:46:14 does it failover back automatically ? 2015-09-08 07:46:21 configure pingu to ping a host or a couple of hosts 2015-09-08 07:46:25 it should 2015-09-08 07:46:37 if it doesnt then it is a bug 2015-09-08 07:47:15 the idea is that pingu pings hosts on internet to test if internet "works" or not 2015-09-08 07:47:27 i have been thinking of adding a dns "ping" 2015-09-08 07:47:33 make it do dns lookups 2015-09-08 07:47:37 in case someone blocks pings 2015-09-08 07:48:48 who maintains the lxc template for alpine? 2015-09-08 07:51:12 ncopa: you do :) 2015-09-08 07:51:17 https://github.com/lxc/lxc/blob/master/templates/lxc-alpine.in 2015-09-08 07:56:04 so, the user is making a container with a template called alpine-latest. where does that come from? the lxc-template apk only has a template called alpine 2015-09-08 08:06:52 it is probably from latest lxc release 2015-09-08 08:39:20 mind if i respond to #4617? 2015-09-08 10:52:41 does someone here have an docker with irssi only on it. somehow i dont understand this concept at all.. 2015-09-08 10:53:22 crow: you and me both. lxc all the way. 2015-09-08 10:53:38 <_ikke_> crow: what concept? 2015-09-08 10:57:26 _ikke_: the "lets embed the entire ubuntu distro in a container but not run init, just our app" concept 2015-09-08 10:57:32 last week some dude here wanted to test a network load and asked for test data to come from 120 different IP addresses. it was childs play with LXC + dhcp + minor edit to the alpine lxc template 2015-09-08 11:00:09 the longest task was for the script to stop 120 containers. about 5 seconds per container 2015-09-08 11:03:20 ScrumpyJack can you show me docker for irssi only (with FiSH plugin would be greate but i could maybe do it myself). 2015-09-08 11:03:24 uggedal well docker. 2015-09-08 11:03:59 crow: nope, as I don't use docker :( 2015-09-08 11:04:56 ScrumpyJack ok, for lxc it should be not that hard, create it, start it, install irssi, profix. 2015-09-08 11:04:58 profit* 2015-09-08 11:08:23 crow: here is a lxc template for alpine running irssi http://st.ilet.to/lxc-irssi 2015-09-08 11:10:43 thanks let me check if i can deploy it correctly, need to take my notes 2015-09-08 11:11:34 you'll need a config file 2015-09-08 11:12:22 do you currently use LXC? (do you have network bridging set up) 2015-09-08 11:31:13 crow: if case you are unfamiliar with LXC http://wiki.alpinelinux.org/wiki/LXC 2015-09-08 11:33:55 ScrumpyJack: would be nice to have the download switch documented on the wiki 2015-09-08 11:36:32 clandmeter: in what context? 2015-09-08 11:37:08 when you want to setup a guest based on ubuntu/debian/... 2015-09-08 11:37:23 with LXC? 2015-09-08 11:37:50 yes 2015-09-08 11:38:22 i didnt even know it existed, and tried to build it myslef 2015-09-08 11:38:34 when just downloading the img is much easier and faster imho 2015-09-08 11:38:39 if you trust it 2015-09-08 11:38:46 ScrumpyJack i know some basic, but before deploying how big is this container? i have currently only 53.6M free ... 2015-09-08 11:39:51 crow: basic alpine takes up 7.4M here 2015-09-08 11:40:42 clandmeter: i wasn't aware of an LXC image. what is that? you mean like an iso? 2015-09-08 11:44:54 ScrumpyJack can it just be deployed like this: sudo lxc-create -n irssi -t ~/lxc/lxc-irssi as i am getting: lxc-create: utils.c: get_template_path: 1320 No such file or directory - bad template: /home/crow/lxc/lxc-irssi but file exist ls /home/crow/lxc/lxc-irssi 2015-09-08 11:46:10 give it the full path 2015-09-08 11:46:24 sudo lxc-create -n irssi -t /home/crow/lxc/lxc-irssi 2015-09-08 11:47:34 its same http://sprunge.us/Vgce 2015-09-08 11:52:27 hmm, the errors are a bit lacking 2015-09-08 11:52:59 what's in your default config file? /etc/lxc/default.conf 2015-09-08 11:58:06 not that much http://sprunge.us/bUDL 2015-09-08 12:05:12 hi when creating a vmware vm for alpine should I choose "other 2.6.x linux (64bit)" as the OS? 2015-09-08 12:08:57 hydrajump i am choosing the: Other Linux 3.x kernel 64-bit 2015-09-08 12:09:09 alpinelinux is long time not on 2.6.x kernel 2015-09-08 12:09:22 crow: great thank you! 2015-09-08 12:09:47 crow: oh did you use the standard iso? 2015-09-08 12:10:18 vanille, or grsec an are both on 3.18.xx something 2015-09-08 12:10:48 ok just didn't quite understand the difference between vanilla and standard 2015-09-08 12:11:20 standard is the grsec one, vanilla withous grsec 2015-09-08 12:12:38 ok 2015-09-08 12:16:56 crow: in short, you need to do something like this first http://st.ilet.to/bridge 2015-09-08 12:22:23 looks like during setup-alpine at the choose fastest mirror it's stuck 2015-09-08 12:22:52 now it's ok again ;) 2015-09-08 12:24:23 ScrumpyJack i am on vanilla so no iptables and such. but i could try rest 2015-09-08 12:26:40 install was super quick compared to ubuntu. Now I just need to find my way around. 2015-09-08 12:30:29 ScrumpyJack that bridge file should be put in /etc/lxc/ or should be added t /etc/lxc/default.conf ? 2015-09-08 12:30:58 i already have an lxc container for build env, but it is using second lan interface not the bridge 2015-09-08 12:45:10 in that case you'll need a bridge for your new container 2015-09-08 12:45:42 the bridge file is a shell script that you should run once 2015-09-08 12:46:05 change INTERNET= to the interface facing outwards 2015-09-08 12:54:31 ScrumpyJack: https://help.ubuntu.com/lts/serverguide/lxc.html#lxc-templates 2015-09-08 12:56:50 hello 2015-09-08 12:57:32 clandmeter: ha yes, very cool 2015-09-08 12:59:26 is it possible to redirect from sys/poll.h to poll.h silently, without warning? the warning breaks things when building with -Werror 2015-09-08 12:59:49 http://slexy.org/view/s2YjTaBz6m 2015-09-08 13:06:45 clandmeter: for something like this? http://tpaste.us/3J6l 2015-09-08 13:08:24 ScrumpyJack: ? 2015-09-08 13:08:46 you were talking about the -- switch right? 2015-09-08 13:09:01 no 2015-09-08 13:09:02 :) 2015-09-08 13:09:06 ok :) 2015-09-08 13:09:28 i was talking about download template 2015-09-08 13:10:41 nevermind, I realised it's an upstream behaviour inherited from musllibc & not something implemented specifically on Alpine 2015-09-08 13:12:29 clandmeter: ok i get you 2015-09-08 13:12:45 so we should build an alpine image? :) 2015-09-08 13:14:32 i'm not sure what the advantage is. what i like about lxc is the lack of images :) 2015-09-08 13:18:07 ScrumpyJack: for alpine images it doesnt matter 2015-09-08 13:18:19 but if you try to create an ubuntu container, it takes some time... 2015-09-08 13:21:16 just timing a debian container 2015-09-08 13:21:46 (with a slow internet connection) 2015-09-08 13:22:59 i wouldn't use lxc without alpine as a guest anyway :) 2015-09-08 13:40:56 real 8m 33.93s 2015-09-08 13:50:24 that was the time it took to build a debian container 2015-09-08 13:50:30 real 0m 12.01s 2015-09-08 13:50:42 that was the time it took to build a alpine container :) 2015-09-08 14:54:01 https://twitter.com/grsecurity/status/641222088246829056 ummmm? 2015-09-08 14:56:41 that's a known issue https://grsecurity.net/announce.php 2015-09-08 15:02:05 and does anyone know how we could still get access for the distro? 2015-09-08 15:02:30 or talk with him? 2015-09-08 15:07:56 darkfader: unstable is still there and can be used. 2015-09-08 16:07:11 so where is libtinfo squirreled away in, it doesn't show up in any ncurses apk that I can find 2015-09-08 18:03:35 is it right that after an alpine install there's only one user, e.g. root? Not like ubuntu where a user is created during install? 2015-09-08 18:04:35 <_ikke_> correct 2015-09-08 18:04:55 <_ikke_> Creating additional users is trivial 2015-09-08 18:05:22 is it the same as on ubuntu with useradd? 2015-09-08 18:05:43 <_ikke_> I think it's adduser 2015-09-08 18:07:15 all i do is adduser -D someuser && addgroup someuser somegroup 2015-09-08 18:07:27 i forget what it is to do both at once but that works 2015-09-08 18:08:16 ok thanks 2015-09-08 18:08:48 another question. I did `apk update` to get update the apk database 2015-09-08 18:09:28 when I ran `apk version -v` I saw lots of packages that had updates. But the version numbers between the installed and available packages are the same 2015-09-08 18:09:44 I ran `apk upgrade` and `apk version -v` still shows the same info 2015-09-08 19:40:48 Ok I solved the above problem 2015-09-08 19:41:21 I just added bind with `apk add bind` but there doesnt' seem to be an entry in /etc/init.d/bind why is that and how to I restart bind? 2015-09-08 19:42:09 <_ikke_> hydrajump: rc-service 2015-09-08 19:44:53 _ikke_: `rc-service bind` ? 2015-09-08 19:45:05 returns rc-service: service `bind' does not exist 2015-09-08 19:45:49 <_ikke_> hydrajump: might it be named? 2015-09-08 19:46:48 _ikke_: awesome thank you! 2015-09-08 19:47:19 _ikke_: is there any difference between running `/etc/init.d/named status` and rc-service named status 2015-09-08 19:47:29 the latter just a shortcut 2015-09-08 19:48:05 <_ikke_> I'm not sure if there is a very big difference 2015-09-08 19:48:12 ok 2015-09-08 19:58:02 so I can start named manually but to have it start on boot according tot he docs here http://wiki.alpinelinux.org/wiki/Comparison_with_other_distros#Runlevel_management 2015-09-08 19:58:44 I should execute `rc-update add -k named default` but I get an error that `k` is an unrecognised option 2015-09-08 20:02:06 <_ikke_> hmm, -k should not be necessary 2015-09-08 20:02:28 looks like the -k option is no longer available. I got it to work with rc-update add named default 2015-09-08 20:05:44 hydrajump: feel free to update the wiki when you find outdated information :) 2015-09-08 20:09:22 hi uggedal ok do I need to create an account to do that? 2015-09-08 20:21:12 hydrajump: yes 2015-09-08 20:39:13 has anyone setup split dns on alpine linux using bind/named? I have a working config working on an ubuntu server, but I'm having trouble trying to make it work on alpine and I'm not sure what if anything is different between the two OSes 2015-09-08 20:40:56 I've setup /etc/network/interfaces to a static ip, /etc/resolv.conf has a nameserver pointing to the local ip 2015-09-08 20:41:39 I've also copied the /etc/bind/named.conf.recursive and uncommented forwarders 2015-09-08 20:55:32 does anyone know where the svg file of the Alpine Linux log is? 2015-09-08 20:56:01 3d printing some stuff and I would like to put it on the item 2015-09-08 21:05:05 systmkor: http://git.alpinelinux.org/cgit/alpine-mksite/tree/_static 2015-09-08 21:06:35 looks like the bind config files that are installed by alpine look quite different from ubuntu. I don't see /etc/bind/named.conf.authoritive and .recursive on ubuntu..hmm 2015-09-08 21:09:20 uggedal: thank you 2015-09-08 21:56:06 Hallo 2015-09-09 00:07:59 hydrajump: which version of ubuntu? 2015-09-09 00:08:37 iirc newer version of ubuntu kind of mangle old-style networking. stuff isn't setup in /etc/resolv.conf anymore on it, everything is handled in systemd scripts 2015-09-09 00:08:48 or /etc/network/interfaces 2015-09-09 00:08:59 unless you force old-style network naming and force-remove networkmanager 2015-09-09 00:27:53 how long before we get node 4.0 on edge? 2015-09-09 00:29:31 ut with needs error warnings 2015-09-09 00:29:53 it works with the glibc hack people are using for java 2015-09-09 03:04:43 So.... the laptop I have only boots from floppy and hdd. So, that leaves me with bg-rescue (which is linux 2.6.32 + uclibc + busybox on 3 disks) as the best option I have seen so far... I am able to mount a usb drive with it and it has ext4 support in the kernel. But not support for my USB USB. 2015-09-09 03:05:11 So, what I want to do now... is download the static apk, and install using the packages from cdrom is that possible? 2015-09-09 03:08:00 Also if anyone has an alternative ... im all ears. I'd like to keep it simple though setting up PXE is a pain .... not what I consider fun after work. If that won't work I might be able to roll a toybox + linux floppy... I'd just have to sort out the boot scripts which I'd probably borrot from bg-rescue since they are pretty nice (handles loading kernel, initramfs and modules form separate floppies... I believe it uses a stub initr 2015-09-09 04:45:15 cb88: If you have access to a SATA to USB adapter, you could take out your hard disk and then from another computer prepare it in the same way as you would prepare a USB stick for installing Alpine. Then you put the hard disk back into your computer. As it runs from RAM, you can make a proper install with setup-alpine. 2015-09-09 04:48:55 engblom: it's a laptop ide drive 2015-09-09 04:49:20 acutally I suppose I could slap it in one of the other ide based laptops I have and install it on there.. 2015-09-09 04:49:36 cb88: That would work 2015-09-09 04:50:45 heh... I just find installing binary packages rather than using a stage3 ala gentoo/funtoo a bit inflexible :D ... but yeah thanks for the suggestion. 2015-09-09 04:52:52 Coming from OpenBSD I understand what you mean. OpenBSD has the base system as a bunch of tgz files. If you could not use the installer for some strange reason you could just extract the tgz files you want, chroot into it and run installboot. 2015-09-09 05:27:29 engblom: yep... I have OpenBSD on my SparcBook 2015-09-09 06:09:56 atomi: as nodejs 4.x has breaking changes maybe we should support both versions? 2015-09-09 06:10:28 nodejs 4.x will be LTS, so maybe its wise to upgrade to it 2015-09-09 09:41:35 anyone using Amazon web services? 2015-09-09 09:56:16 with Alpine Linux, obviously :) 2015-09-09 10:09:16 i had AMI image of alpine earlier 2015-09-09 10:09:47 converted using their scripts? 2015-09-09 10:10:49 no, built using setup-bootable from non-alpine amazon vm 2015-09-09 10:10:53 iirc 2015-09-09 10:10:56 it's long ago 2015-09-09 10:11:19 the initial setup was manual; and ever since i just used setup-bootable / update-kernel 2015-09-09 10:11:24 from the alpine vm 2015-09-09 10:11:59 we should have a build site. builds various images for various VM techs ready for hosting :) 2015-09-09 10:39:35 anyone experiencing vim periodically becoming unresponsive for a few seconds? 2015-09-09 10:42:39 uggedal: that's the idiotic autosave on a slow sd card 2015-09-09 10:43:22 it does a 'sync' when it updates the vim .swp file etc 2015-09-09 10:43:31 and that means a blocking disk flush 2015-09-09 10:45:19 darkfader: this is on a ssd... 2015-09-09 10:45:25 meh 2015-09-09 10:48:02 same result with vim -N -u NONE 2015-09-09 10:53:47 no idea then, that's odd 2015-09-09 10:54:04 oh and is your ~ also on the ssd? 2015-09-09 10:55:50 darkfader: yes, everything on ssd 2015-09-09 10:56:44 just tested, same thing happens on void linux with musl 2015-09-09 10:56:59 not on same version on void linux with glibc 2015-09-09 11:01:26 /usr/bin/vim is owned by vim-7.4.712-r0 2015-09-09 11:01:31 no problems here 2015-09-09 11:02:06 hmm, nvi has the same problem but not as frequent 2015-09-09 11:03:08 musl handles flushes to stdout slightly different. 2015-09-09 11:03:14 uggedal: can you please look at slabtop and /proc/meminfo (dirty) during the hangs 2015-09-09 11:03:18 I remember a patch for some of the coreutils. 2015-09-09 11:03:23 Gottox: oh cool 2015-09-09 11:04:09 https://github.com/voidlinux/void-packages/blob/master/srcpkgs/util-linux/patches/fix-musl-2.patch 2015-09-09 11:04:27 was util-linux 2015-09-09 11:04:49 but vim could have a similiar issue. 2015-09-09 11:05:09 (or ncurses, no idea) 2015-09-09 11:45:30 Sleaker: ubuntu 10.04 but it looks like the bind version running there is 9.4.2 and alpine has 9.10.2 so things have probably changed quite a bit 2015-09-09 11:50:31 Hi everyone 2015-09-09 11:52:06 Does somebody knows why there is a user named "operator" ? 2015-09-09 11:52:23 <_mjones> noknook: ehlo 2015-09-09 11:52:55 noknook: where? 2015-09-09 11:53:20 <_mjones> noknook: I don't know specifically. Modern practise is to install separate users for running application daemons. 2015-09-09 11:53:45 ah ok 2015-09-09 11:54:57 <_mjones> noknook: I assume you see 'operator' user on a base install? 2015-09-09 11:58:28 yes, it's from a base install, in etc/passwd 2015-09-09 11:58:32 <_mjones> Yes, right here after uucp. Ah, tradition. 2015-09-09 11:59:20 <_mjones> uucp is so obsolete I might start using it again because it's so vintage. 2015-09-09 12:00:50 wait till it's added to systemd 2015-09-09 12:01:46 <_mjones> darkfader: comment of the week. ;) 2015-09-09 12:02:51 uucp for devops could also be a good meme 2015-09-09 12:04:07 <_mjones> That sounds like a talk I should give on April 1st. 2015-09-09 12:05:42 it could be a great thing 2015-09-09 12:06:22 <_mjones> noknook: the operator account is for the operator who mounts your tapes and whatnot. 2015-09-09 12:07:08 <_mjones> Get me a scratch tape on that TK50, stat! 2015-09-09 12:09:35 _mjones: thanks :) it still there because no one changed this file since 1981, am I right ? The fact that he had /bin/sh as a shell isn't a security issue? 2015-09-09 12:11:17 <_mjones> noknook: operator has no password in /etc/shadow. I doubt there's a reason why /bin/sh is needed in practise, but if you change it you should keep an eye for any breakage, although I doubt there will be any. 2015-09-09 12:13:15 <_mjones> noknock: Alpine uses an unconventional single-binary userland with an alternative libc on a nonderivative posix implementation compiled with an unusual set of security patches, and you think the /etc/password is from 1981? ;) 2015-09-09 12:14:42 <_mjones> I bet that passwd file comes from one of the base components in Linux From Scratch. I'm now curious which base component, and its derivation. 2015-09-09 12:20:45 Ok, let me give you a big thanks since your answers is polite, fun, educationnal and complex in the same time :D 2015-09-09 12:21:18 But "one more thing" 2015-09-09 12:22:45 I don't understand something. If it has no password in etc/shadow, you cannot login as user. But at first boot, root had no password too. How is it done? 2015-09-09 12:22:58 <_mjones> noknook: now you're making me feel guilty that I couldn't give a real, definitive answer. 2015-09-09 12:24:20 <_mjones> noknook: the second field (separate by :) in /etc/shadow is the password field. Putting a "!" character in that field means the password will never match, and effectively locks the account from logins. 2015-09-09 12:26:12 <_mjones> So if the /etc/shadow says: username:!: or username:*: then you can't match the password. You can still su to the user, though. 2015-09-09 12:28:06 I definitely had to take a look at passwd man. Thanks so far, keeping this chan in fav 2015-09-09 16:38:15 hi what's the equivalent of this on ubuntu `host $(hostname)` on alpine? 2015-09-09 16:38:37 do I need to install a package or is there another toll already installed similar to host? 2015-09-09 16:38:56 run setup-hostname 2015-09-09 16:40:51 <_mjones> host does dns lookups 2015-09-09 16:41:15 ScrumpyJack: I don't want to setup the hostname. I'd like to verify it. The output of `host $hostname` looks like this: 2015-09-09 16:41:15 ah 2015-09-09 16:41:18 <_mjones> hydrajump: dig returns the same information, but lots more besides 2015-09-09 16:41:35 mail.yourdomain.com has address 192.168.1.30 2015-09-09 16:41:44 <_mjones> hydrajump: uh, let me look and see if it's in bind-utils or what package provides. 2015-09-09 16:41:48 _mjones: I installed drill 2015-09-09 16:42:02 which works great doing dig like lookups 2015-09-09 16:42:23 <_mjones> ACTION tries to remember the damn 'yum whatprovides' equivalent on debian, to check a non-Alpine machine. 2015-09-09 16:42:25 I was just following a tutorial on split dns elsewhere and that host command was shown so I thought I'd see if it worked on alpine 2015-09-09 16:42:43 _mjones: let me see if I remember 2015-09-09 16:42:46 <_mjones> It should be a matter of what package provides it. 2015-09-09 16:43:00 hydrajump: host `cat /etc/hostname` ? 2015-09-09 16:43:18 _mjones: maybe apt-cache search host or something 2015-09-09 16:43:40 /usr/bin/host is owned by bind-tools-9.10.2_p3-r1 2015-09-09 16:43:51 <_mjones> Well that does it forward. Which is probably good enough. 2015-09-09 16:44:19 weird when I do `which host` it says not found 2015-09-09 16:44:28 apk add bind-tools 2015-09-09 16:45:07 ScrumpyJack: thank you ;) 2015-09-09 16:45:42 how could I have found whih package contained host? 2015-09-09 16:45:53 apk info --who-owns /usr/bin/host 2015-09-09 16:45:53 apk search host just returns packages named host 2015-09-09 16:45:58 <_mjones> So, 'apk add bind-tools'. Same package name on Alpine. 2015-09-09 16:45:59 :D 2015-09-09 16:46:06 ScrumpyJack: thank you 2015-09-09 16:46:22 and you _mjones 2015-09-09 16:46:32 --who-own that only works if you have it installed 2015-09-09 16:46:33 I'm really liking alpine so far 2015-09-09 16:46:44 otherwise you can search the package repo 2015-09-09 16:46:53 ACTION digs out the link 2015-09-09 16:46:59 <_mjones> typically search will find what you want. But sometimes you just need to know the name. Same on other distributions. 2015-09-09 16:47:03 https://pkgs.alpinelinux.org/packages 2015-09-09 16:47:47 like this https://pkgs.alpinelinux.org/contents?filename=host&path=&pkgname=&repo=all&arch=x86_64 2015-09-09 16:47:57 looks like i have a problem with my bind setup as host $(hostname) returns `Host mail not found: 3(NXDOMAIN) 2015-09-09 16:47:58 easily scitpable 2015-09-09 16:48:07 <_mjones> Alpine package names are usually similar to debian or arch, so sometimes a quick bing will let you know what you need to install. 2015-09-09 16:48:17 but drill example.com mx returns the correct info 2015-09-09 16:48:38 thanks for the tips 2015-09-09 16:48:41 <_mjones> hydrajump: Is it inside your split-horizon or external? 2015-09-09 16:49:04 _mjones: is what inside? 2015-09-09 16:49:17 the mail server? 2015-09-09 16:50:49 could it be because i'm missing `search example.com` in /etc/resolv.conf? 2015-09-09 16:51:10 yep! 2015-09-09 16:54:44 <_mjones> Always try to use FQDNs everywhere. 2015-09-09 16:55:16 <_mjones> I think the 'domains' keyword isn't even supported on musl, and thus not on Alpine. 2015-09-09 16:56:15 <_mjones> TLS and x509 certs mean we can't do a few cool things with nonqualified hostnames anymore, viz. http://printing 2015-09-09 16:57:45 heya 2015-09-09 16:57:58 does Virtualbox run on alpine 2015-09-09 17:06:38 it might need glibc or something :/ 2015-09-09 17:14:37 <_mjones> It's open source, so...try to give it a quick compile? 2015-09-09 17:16:35 <_mjones> In aports tree there are two items: testing/virtualbox-additions-grsec and unmaintained/xf86-video-virtualbox . 2015-09-09 17:18:22 xf86-video-virtualbox would be a guest vm video driver 2015-09-09 17:18:52 that would be necessary for running Xorg on alpine when installed into a virtualbox vm 2015-09-09 17:19:02 not when using alpine as a virtualbox host. 2015-09-09 17:19:24 and the virtualbox-additions-grsec should be the same thing. guest-only additions for use with a grsec kernel 2015-09-09 17:19:29 yeah those are both packages for vbox guests 2015-09-09 17:19:35 i want to use alpine as a host maybe 2015-09-09 17:19:46 those packages don't do anything for the host 2015-09-09 17:19:53 they are specific to the guest-os 2015-09-09 17:20:07 are you planning on running alpine in your guest? 2015-09-09 17:20:13 no host 2015-09-09 17:20:22 then neither of those packages do anything for you. 2015-09-09 17:22:37 <_mjones> I'm not a virtualbox user. I kvm and still do some xen. 2015-09-09 17:23:00 epictetus: basically youd need to obtain virtualbox from oracle: https://www.virtualbox.org/wiki/Linux_Downloads 2015-09-09 17:23:25 I think the only one that might work is the All distributions one. 2015-09-09 17:24:08 if the binaries don't work, sources are available to compile from their download page too 2015-09-09 17:26:18 <^7heo_> did anyone run virtualbox as a host on alpine? 2015-09-09 17:38:12 <_mjones> ^7heo: that's the question. Probably not. ;) 2015-09-09 20:22:09 during the setup-alpine routine I was asked which ntp server to use. I chose openntpd. How can I find where its config files are located? On ubuntu I could use `whereis ...` and it would usually give me the location of the binary as well as any config locations 2015-09-09 20:23:02 hydrajump, https://pkgs.alpinelinux.org/package/main/x86_64/openntpd 2015-09-09 20:23:08 and click on Contents of package 2015-09-09 20:24:16 icicled: ok so there's no `apk` command to get this kind of info? 2015-09-09 20:24:31 there probably is...I'm not at home - can't check >_> 2015-09-09 20:24:52 no worries that link helps too! 2015-09-09 20:28:28 hydrajump: apk info -L openntpd 2015-09-09 20:30:29 thank you uggedal! 2015-09-09 20:48:47 I'm tackling users now and coming from ubuntu server I'm used to a user being created during install and using `sudo` whenever I needed to do something as root. A the moment as I'm learning alpine everything has been done as root including connecting via ssh. THis is obviously not a good security practice. 2015-09-09 20:49:39 I'm looking around the file system and I don't see a /etc/sudoers file. Does one not exist on alpine? 2015-09-09 20:50:04 hydrajump, install sudo 2015-09-09 20:50:07 apk add sudo 2015-09-09 20:50:51 <_mjones> If you like sudo, you can keep it! 2015-09-09 20:51:18 <_mjones> I messed that up. "If you like your sudo, you can keep it!" 2015-09-09 20:51:57 how do aficionado alpine users do it? Do you install sudo? 2015-09-09 20:52:18 Im happy to do things differently here if recommended over what I've done before ;) 2015-09-09 20:52:38 I use it...simply because I'm used to it 2015-09-09 20:54:02 being that alpine is focused on security is there any difference in security between running as the default root user vs sudo _if_ root user is properly protected with a strong passphrase and ssh only uses PKI? 2015-09-09 20:54:32 it means that everything by default runs as root, right? 2015-09-09 20:54:41 all installed packages/ daemons? 2015-09-09 20:56:55 Hello all, trying to install alpine to pi with luks + lvm. I got my partitions, mounted, ran setup-disk /mnt, and regenerated mkinitfs. 2015-09-09 20:57:32 Now I'm trying to run `extlinux` which doesn't seem to exist, nor can I find package for `syslinux` for some reason, could use a little assistance please 2015-09-09 21:00:22 <_mjones> zapadlo: I think the RPi install is stateless. You just write the whole image to the card and boot it, I think? 2015-09-09 21:00:54 <_mjones> icicled, hydrajump: no, no, I certainly don't run as root all the time. I use 'su'. 2015-09-09 21:01:26 _mjones: I'm trying to make it permanent by installing to sdcard 2015-09-09 21:01:31 <_mjones> Specifically I usually run a tmux session (functionally the same as gnu screen if you know that) and su to root in only one or two of the sessions, as needed. 2015-09-09 21:01:51 _mjones: So I effectively boot from usb stick and sdcard is for all terms and purpose my HD 2015-09-09 21:02:30 <_mjones> zapadlo: I think the intention is to write the boot image directly to the sdcard using appropriate adapters. Do you lack adapters, etc.? 2015-09-09 21:02:55 _mjones: is there any other linux distro that by default doesn't have sudo and only sets up a root user? 2015-09-09 21:02:59 Unsure what you mean by adapters, sorry 2015-09-09 21:03:43 <_mjones> hydrajump: uh. Not sure. I have Arch and Debian machines. 2015-09-09 21:03:56 <_mjones> Setting up a user and using or not using sudo are two mostly-separate things. 2015-09-09 21:04:32 <_mjones> Set up a user with 'adduser'. It's not intended to log in and run as root, no. 2015-09-09 21:04:49 <_mjones> Whether you elevate to root with 'su' or 'sudo' is a separate question. 2015-09-09 21:06:24 ok and once a user has been setup, e.g. root and hydrajump then it's a matter of changing the sshd_config to prevent root login and then I'd be in a similar familar state to an ubuntu server? 2015-09-09 21:06:42 <_mjones> Correct, hydrajump. 2015-09-09 21:07:06 _mjones: you're very helpful! 2015-09-09 21:07:08 <_mjones> You've now given me occasion to think about an option to do this from the installer. 2015-09-09 21:07:29 _mjones: you're one of alpine's creators? 2015-09-09 21:07:34 yes please! 2015-09-09 21:07:46 and perhaps disable root login via ssh 2015-09-09 21:07:49 <_mjones> Frankly, I don't think about the installer much in general. And this is so routine for a unix veteran I don't think much about it. But it bears consideration. 2015-09-09 21:08:18 better to use safe defaults than not 2015-09-09 21:08:18 <_mjones> My only recent concern is that there's no stock UEFI boot just yet, although with some luck it will make the 3.3.0 release. 2015-09-09 21:08:31 <_mjones> No, I' 2015-09-09 21:08:55 <_mjones> I'm not a creator, and scarcely even a contributor just yet. 2015-09-09 21:09:26 <_mjones> But I can ask, I can suggest, and I can contribute patches for new behaviour if I want. 2015-09-09 21:09:55 well no matter what I appreciate the help I've been getting here! 2015-09-09 21:10:25 <_mjones> Please bear in mind that 'sudo by default' is a relatively new thing in unix and even in Linux. The first I ever saw of it was Ubuntu. 2015-09-09 21:11:39 <_mjones> The way going back more than 40 years was 'su'. I understand the use cases of sudo, and I use su on my boxes, and only use sudo for custom rulesets in multiuser situations. 2015-09-09 21:11:49 using su or sudo to run things as root is a lot less secure than using ssh 2015-09-09 21:12:19 <_mjones> It's the security defaults out of the box, and user expectations, that bear some renewed consideration. 2015-09-09 21:12:38 <_mjones> dalias: don't confuse people with being technically correct. -Ish. 2015-09-09 21:12:55 especially if you make proper use of pubkeys and the authorized_keys file's forced command field 2015-09-09 21:13:38 dalias: so you ssh as root when you need to elevate instead of using `su/sudo`? 2015-09-09 21:13:41 it's secure because you're actually using the permissions model to do what you want rather than bypassing it with a suid binary then attempting to emulate some version of it 2015-09-09 21:14:07 yeah. but be sure password auth is disabled 2015-09-09 21:14:25 <_mjones> ssh-based rbac and the philosophy of suid wasn't the question. ;) 2015-09-09 21:14:33 interesting. eveything I've seen about ssh server configs shows permitrootlogin as no 2015-09-09 21:15:18 good luck brute-forcing a 2048-bit key :) 2015-09-09 21:15:49 (or, these days, the equivalent in ecdsa if you prefer :) 2015-09-09 21:15:54 <_mjones> key-based logins only is orthogonal with logins as root, also. 2015-09-09 21:16:21 right 2015-09-09 21:16:27 <_mjones> hydrajump: we're talking about....let's say special use-cases, not specific to Alpine at all. 2015-09-09 21:16:37 *nod* none of this is alpine-specific 2015-09-09 21:16:53 yeah that I understand ;) I don't follow your last point though _mjones 2015-09-09 21:17:02 "key-based logins only is orthogonal with logins as root, also." 2015-09-09 21:17:07 <_mjones> The root logins is about suid. I've been familiar with product suidless machines for a long time. 2015-09-09 21:17:21 <_mjones> Thomas Ptacek used to run his public machine that way twenty years ago. 2015-09-09 21:17:24 *nod* 2015-09-09 21:17:42 suid binaries are a high risk item, and completely unnecessary on modern systems 2015-09-09 21:18:03 suid binaries being `sudo` `su`? 2015-09-09 21:18:14 and anything else with the suid (+s) bit set 2015-09-09 21:18:19 ok 2015-09-09 21:18:37 <_mjones> If root can only happen through a direct login session, you don't need su and suid, which represent critical attack surface because they're programs with the ability to hand out root privileges. 2015-09-09 21:18:56 <_mjones> I meant you don't need su and sudo to be suid. 2015-09-09 21:19:24 <_mjones> Now, otoh, this reminds me I need to be all keys, which I've been avoiding. 2015-09-09 21:20:03 <_mjones> Last scale production shop was an AD shop, so our logins and sudo were tied to AD creds and AD groups, respectively. Not key auth based. 2015-09-09 21:21:11 <_mjones> hydrajump: sometimes these techniques are suitable when we have hoardes of undifferentiated VM machines or 'cattle', unlike traditional maintained servers which can correspondingly be dubbed 'pets'. 2015-09-09 21:21:59 <_mjones> I hadn't thought about ssh forced command though. That's an interesting option. 2015-09-09 21:22:16 sure I'm familiar with the pets vs cattle convention 2015-09-09 21:23:48 so dalias say you've ssh'd into a server as dalias and you need to be root to do something like change a config file. You'd then start a second ssh session as root, do the config change and disconnect that session? 2015-09-09 21:25:16 <_mjones> Huh, look at that: alpine-dev:~$ login login: must be suid to work properly 2015-09-09 21:25:27 "rather" from a sense of having a system with smaller attack surface 2015-09-09 21:25:32 it might be less convenient 2015-09-09 21:25:48 _mjones, that's a bug that should be fixed 2015-09-09 21:25:53 login definitely does not need to be suid 2015-09-09 21:25:57 it's invoked as/by root 2015-09-09 21:26:13 probably some busybox stupidity :) 2015-09-09 21:26:46 btw for ssh-to-root you have 2 approaches you can use, or a mix 2015-09-09 21:26:46 <_mjones> The much less common use requires suid. Probably not a use most people want, any more. 2015-09-09 21:27:13 you can have the private key (encrypted with passphrase) on the same host so you can just ssh root@localhost 2015-09-09 21:27:56 or you can keep the private key only on a more-trusted system you originate from 2015-09-09 21:28:15 <_mjones> oh know, now you went and did it. Passphrase on keys! 2015-09-09 21:28:55 <_mjones> Yes, we all have passphrases on our https certs as well. We type it in on console on every machine when we restart the webserver. Don't tell anyone. ') 2015-09-09 21:29:02 there's really no benefit in having separate root/user accounts if the user account has a non-passphrase-protected key that can login as root 2015-09-09 21:29:38 otoh having the non-passphrase-protected private key on a separate trusted system makes some sense 2015-09-09 21:29:42 <_mjones> Accounting and clarity for non-hostile userbase. 2015-09-09 21:30:06 also you can mix passphrase-protected and non-passphrase-protected private keys for different purposes if you use forced command in authorized_keys 2015-09-09 21:30:23 <_mjones> I've been thinking about jumpboxes. Wheel group makes an interesting one called fudo, based on capabilities-limited freebsd, that will log everything for compliance. 2015-09-09 21:31:17 obviously automated processes need non-passphrase-protected priv keys to use 2015-09-09 21:31:26 but they should probably be restricted in what commands they can run 2015-09-09 22:09:53 <^7heo> _mjones: I'm asking; because I tried. 2015-09-09 22:10:20 <^7heo> And found it MUCH easier to install Kvm 2015-09-09 22:11:07 <_mjones> kvm <3. 2015-09-10 03:00:33 when adding a daemon, is there a particular uid, gid I should use? 2015-09-10 03:00:42 I'm setting up smtpd and smtpq users for opensmtpd 2015-09-10 03:00:57 I want to know which gid/uid to add to the pre-install script 2015-09-10 03:01:11 or does it matter? 2015-09-10 05:12:10 joncfoo: uid/gid is the user/group 2015-09-10 05:13:00 from what I've seen the normally prescribed way to do things is to create a new user/group for the service, if it's going to do that with a the gid/uid in the script then there are some guidelines on the web 2015-09-10 05:14:08 this is an example of how arch sets there's up: https://wiki.archlinux.org/index.php/DeveloperWiki:UID_/_GID_Database 2015-09-10 05:52:41 what"s the recommended SIP server for small install (~10 users)? 2015-09-10 06:41:21 <__number5__> asterisk 2015-09-10 09:22:07 hmm, something must have changed 2015-09-10 09:22:18 extending $PATH does not work anymore 2015-09-10 09:22:44 i can extend the variable, but i cannot execute stuff from the extened new path 2015-09-10 09:37:26 in a shell? 2015-09-10 09:37:38 yes 2015-09-10 09:37:56 usually i added /home/leo/repos/leo/scripts to $PATH 2015-09-10 09:38:01 and then i could execute scripts in there 2015-09-10 09:38:16 but now even after extending $PATH they are not found :/ 2015-09-10 09:38:17 i'll try here 2015-09-10 09:40:13 works for me in zsh 2015-09-10 10:09:35 hmm, very strange 2015-09-10 10:11:00 hi guys, anyone recommend a FTP server? 2015-09-10 10:16:39 alpine has packages for proftpd, pute-ftpd and vsftpd 2015-09-10 11:03:06 what you ppl use for monitoring cpu load (maybe to writhe a graphs to) 2015-09-10 11:14:06 conky :) 2015-09-10 11:15:53 ScrumpyJack, +1 2015-09-10 11:18:27 here is the top left had corner of conky output on my alpine linux desktop showing CPU usage http://st.ilet.to/conky.png 2015-09-10 11:43:02 hi after adding dovecot I saw this during install `unable to write 'random state'` is that something to be worried about? 2015-09-10 11:43:09 is it an alpine issue or dovecot issue? 2015-09-10 12:01:04 looks like it may be openssl related 2015-09-10 12:11:41 ScrumpyJack: hmm, when i searched earlier for ftpd, only vsftpd showed up ! hmph ! 2015-09-10 12:12:42 wierd 2015-09-10 12:23:19 thanks for the info Sleaker, just what I needed! 2015-09-10 12:30:14 I've ended up with http://sprunge.us/ODPH 2015-09-10 12:33:17 is there a way to add a user & set the password programatically 2015-09-10 12:37:49 hendry: adduser & passwd? 2015-09-10 12:38:54 joncfoo: i see it has a -d switch 2015-09-10 12:40:03 joncfoo: -d switch doesn't work ... 2015-09-10 12:40:21 it just sets it to empty 2015-09-10 12:47:49 arghh, why does alpine have SO MANY users by default http://s.natalian.org/2015-09-10/1441889254_1144x1036.png 2015-09-10 12:50:36 hendry: it's -D 2015-09-10 12:50:38 caps 2015-09-10 12:50:54 err thats for adduser 2015-09-10 12:50:56 nvm 2015-09-10 12:55:59 hendry: it's all relative. i think that's not a lot. lots could be removed. is that alpine or alpine-mini? 2015-09-10 12:58:18 ScrumpyJack: alpine 2015-09-10 13:12:12 strange I ran `halt` which according to the wiki is equivalent to shutdown -h now, but in the console I see 5 asterisks and it is still powered on 2015-09-10 13:12:46 ncopa: How's the situation with grsecurity? 2015-09-10 13:15:28 slashbeast: no response from spender yet 2015-09-10 13:15:36 ncopa: okey 2015-09-10 13:17:27 poweroff works but halt doesn't 2015-09-10 13:19:58 ncopa: please let me know once you will have any update about it 2015-09-10 13:30:42 joncfoo: usermod in ubuntu lets you add pre-hashed passwords to users. 2015-09-10 13:30:51 not sure if that's available in alpine 2015-09-10 13:31:28 otherwise you can just sed in the password into /etc/shadow 2015-09-10 13:31:39 but it does need to be pre-hashed 2015-09-10 13:41:40 Does the device support acpi? 2015-09-10 13:42:06 *blink* 2015-09-10 13:42:32 sry too late, didn't see that the conversation moved :( 2015-09-10 13:42:47 But one question 2015-09-10 13:43:08 I had to use GeoIP2, which is not included in alpine packages 2015-09-10 13:43:13 WHat can I do? 2015-09-10 13:43:23 hendry: looks like you've installed stuff there. 2015-09-10 13:45:07 although I haven't put apache, cyrus, mailer, ftp, or mysql on my system and the users are there anyhow 2015-09-10 13:45:09 odd 2015-09-10 13:45:22 maybe the prebuilt image has them all 2015-09-10 13:48:16 joncfoo: ah found it: # echo "linuxpassword" | passwd --stdin linuxuser 2015-09-10 13:49:41 Sleaker: chpasswd 2015-09-10 13:50:07 oh cool didn't know about that tool 2015-09-10 13:51:06 noknook: geoip is in edge i believe 2015-09-10 14:18:14 Nice, found kamailio-geoip2-4.3.1-r0.apk in edge. Does packets in 'edge' will be released in the 'main' repo of the next version? 2015-09-10 14:21:50 noknook: yes 2015-09-10 14:22:06 edge/main is what will become v3.3/main 2015-09-10 14:26:36 ncopa: apk update && apk upgrade will upgrade the kernel is available? 2015-09-10 14:34:14 ScrumpyJack: yes it should 2015-09-10 14:34:27 if you have disk install that is 2015-09-10 15:07:26 what about the RPi/arm install? 2015-09-10 18:01:45 why does glib-dev depend on expat-doc? 2015-09-10 18:38:41 <^7heo> okay, so 2015-09-10 18:39:16 <^7heo> I have manually patched the /etc/init.d/networking script to boot faster (it basically detects if the interface is up and doesn't dhcpc it if it isn't) 2015-09-10 18:39:23 <^7heo> and now udev starts without any problem 2015-09-10 18:39:30 <^7heo> also it boots like 10x faster. 2015-09-10 18:40:16 <^7heo> I still have an issue here, I believe that I might have done it wrong somehow, because I boot without IP, I need to restart /etc/init.d/networking in order to get an IP 2015-09-10 18:40:30 <^7heo> ncopa: ^ 2015-09-10 18:40:53 <^7heo> (correction: it detects if the interface is down and doesn't do anything if it is) 2015-09-10 18:46:43 <^7heo> okay, it seems that at the moment when I do the detection, is_down is always true. 2015-09-10 18:46:48 <^7heo> sucks. 2015-09-10 18:47:04 <^7heo> but somehow it prevents udev from failing to start, which is interesting. 2015-09-10 21:05:25 dalias: since glibc-dev -> gettext-dev -> expat-dev -> expat-doc 2015-09-10 21:05:41 dalias: don't understand why it happens though 2015-09-10 21:06:06 dev packages should not depend on their doc packages; that seems like a bug in the package 2015-09-10 21:06:34 dalias: yeah, it's a bug. the thing is I don't understand why the dep is added 2015-09-10 21:13:55 dalias: are you on edge version? 2015-09-10 21:14:26 pushed a rebuild and the dep is gone 2015-09-10 21:14:37 my guess is a bug/feature in previous abuild version 2015-09-10 21:16:03 :) 2015-09-10 23:28:30 I'm creating a new aport and I get a message when doing abuild: "You should not have $install in source" 2015-09-10 23:28:38 I'm following what's on the wiki 2015-09-10 23:29:00 and if I don't put $install in source then the pre-install script doesn't get included 2015-09-10 23:29:05 am I missing something? 2015-09-10 23:30:48 anyone seen this issue on alpine https://bugs.archlinux.org/task/43789 2015-09-10 23:30:52 I dont think they ever get included 2015-09-10 23:30:59 i dont remember the reason 2015-09-10 23:31:11 I'm getting that error msg now "postmap: warning: smtputf8_enable is true, but EAI support is not compiled in 2015-09-10 23:31:14 " 2015-09-10 23:35:20 clandmeter: I just tested it an the pre-install script I have does get executed 2015-09-10 23:35:38 yes 2015-09-10 23:35:47 they get pulled in automagically 2015-09-10 23:35:53 oh I see 2015-09-10 23:35:54 but they are not added to checisums 2015-09-11 01:29:26 just submitted an aport for opensmtpd 2015-09-11 01:29:31 fingers crossed =] 2015-09-11 14:10:11 Is this an issue for alpine linux ? http://grsecurity.net/announce.php 2015-09-11 14:18:42 <_mjones> noknook: Yes, but a solution is in progress. As of this morning, it's looking like there's probably going to be a good resolution. 2015-09-11 14:19:40 <_mjones> Even if not, it's not a showstopper. It would just mean there would need to be a significant effort to accomodate this by the Alpine Linux project. 2015-09-11 14:28:28 <_ikke_> Any way to find out why alpine hangs when booting from USB at booting syslinux or something like that? 2015-09-11 14:30:15 _mjones: Thanks, do you have more informations about this resolution? 2015-09-11 14:34:47 noknook: we already use a 3rd party port of the patch 2015-09-11 14:35:02 since we use the unsupported 3.18 kernel 2015-09-11 14:35:17 whcih means we already do some porting work 2015-09-11 14:35:37 we can continue do that with grsecurity testing patches 2015-09-11 14:35:42 but it means more work 2015-09-11 14:35:52 i have written to spender to ask for advice 2015-09-11 14:36:11 we dont think his intention was to hit opensource projects 2015-09-11 14:37:09 so we hope we can get access to changelog 2015-09-11 15:06:55 ncopa, do you know much about the gcc esp patch? 2015-09-11 15:07:12 i have a simple patch in my toolchain adding self specs to make -pie default 2015-09-11 15:07:18 but i can't get the patch to add -nopie to work 2015-09-11 15:07:29 it always says unrecognized option :( 2015-09-11 15:08:26 <_ikke_> I'm trying to boot a system with qemu to test it, but the syslinux message keeps appearing. Anyway to get more information? 2015-09-11 15:09:03 dalias: there was long time since i messed with the hardened toolchain 2015-09-11 15:10:21 dalias: maybe this? http://git.alpinelinux.org/cgit/aports/tree/main/gcc/116_all_gcc47_nopie_option.patch 2015-09-11 15:10:40 that's what i'm doing but it doesn't work 2015-09-11 15:11:13 iirc there used to be some patching of the specs 2015-09-11 15:11:23 <_mjones> _ikke_: What message? Is the qemu x86_64? 2015-09-11 15:11:26 maybe i just need a clean build (broken deps?) 2015-09-11 15:12:01 there are some nopie specs here: http://git.alpinelinux.org/cgit/aports/tree/main/gcc/140_all_gcc49_config_esp.patch 2015-09-11 15:12:07 i dunno 2015-09-11 15:12:10 i need to go now 2015-09-11 15:12:14 have a nice weekend 2015-09-11 15:12:23 <_ikke_> _mjones: qemu-system-x86_64 2015-09-11 15:12:32 <_ikke_> _mjones: It just keeps hanging 2015-09-11 15:12:38 ok thanks 2015-09-11 15:12:41 have a nice weekend 2015-09-11 15:15:03 <_mjones> _ikke_: are you invoking qemu through libvirt or directly? If you start it directly it's likely you can get any error messages issued by qemu. 2015-09-11 15:16:28 <_mjones> _ikke_: you're booting the installer media, correct? 2015-09-11 15:16:30 <_ikke_> _mjones: directly 2015-09-11 15:16:42 <_ikke_> This was a running system, but somehow, it won't boot anymore 2015-09-11 15:16:46 <_ikke_> usb install 2015-09-11 15:17:36 <_mjones> _ikke_: ah. After a kernel upgrade, perchance? 2015-09-11 15:18:01 <_ikke_> _mjones: no 2015-09-11 15:18:30 <_ikke_> _mjones: We didn't touch the system 2015-09-11 15:18:33 <_ikke_> We shutdown the blade 2015-09-11 15:18:44 <_ikke_> and that specific blade wouldn't boot anymore, so the blade was changed 2015-09-11 15:18:50 <_ikke_> And now it's hanging at boot, even with qemu 2015-09-11 15:20:11 <_ikke_> https://imgur.com/aB3PBhh 2015-09-11 15:20:41 <_ikke_> I boot it just with qemu, just to test, no libvirt involved 2015-09-11 15:21:40 <_ikke_> https://gist.github.com/anonymous/105597d606e2f20633a4 2015-09-11 15:21:45 <_ikke_> this is the extlinux.conf file 2015-09-11 15:45:45 <_ikke_> How would I restore initramfs and vmlinuz? 2015-09-11 15:47:19 <_ikke_> boot partition has been corrupted (fsck found several problems) 2015-09-11 17:57:30 quick question - does alpine linux run well on the latest lenovo x1 carbon (3rd gen) 2015-09-11 17:58:01 i see a "known working configuration" Lenovo X1 Carbon (1st gen) 2015-09-11 17:59:11 it looks like that piece is about the "modern Intel GPU chipsets covered by the i915 driver" support 2015-09-11 18:51:16 <_mjones> "On Ubuntu 14.04, you can now simply do apt-get install zram-config." Oh, that's nice. 2015-09-11 19:15:22 hey there 2015-09-11 19:15:36 I am trying to install unzip on my docker alpine container 2015-09-11 19:15:51 but I am getting unzip(missing) 2015-09-11 19:18:26 nevermind 2015-09-11 19:21:33 <_mjones> jmonreal: what ended up being the problem? 2015-09-11 19:27:38 it was already installed 2015-09-11 19:28:04 <_ikke_> _mjones: Turned out the boot partition probably got corrupted 2015-09-11 19:28:15 <_ikke_> _mjones: fsck returned all kinds of problems 2015-09-11 19:28:37 <_mjones> danke, both. 2015-09-11 19:31:04 <_ikke_> _mjones: I probably got the bootloader fixed, but I suspected the initramfs / vmlinuz-grsec files got corrupted. How would I restore those? It was a very old alpine setup. 2015-09-11 19:32:20 <_mjones> _ikki_: If it's important, take a backup first. Then I would run mkinitfs and see if it can regenerate the initrd for you. 2015-09-11 19:33:16 <_mjones> Also, if you're up and running and it's old, an 'apk --update upgrade' would likely pull in a new kernel, obviating any existing kernel corruption. 2015-09-11 19:34:36 <_ikke_> _mjones: right, but that would require me somehow to get the system running in the first place 2015-09-11 19:34:55 <_ikke_> probably chrooting it from an existing alpine system 2015-09-11 19:35:03 <_mjones> Oh! I thought you had it running. Sorry. 2015-09-11 19:35:21 <_ikke_> No, it wouldn't boot yet 2015-09-11 19:35:38 <_ikke_> We decided to reinstall the system 2015-09-11 19:37:01 <_mjones> Good idea, where possible. 2015-09-11 19:37:12 <_mjones> Any idea on the root cause of the corruption? 2015-09-11 19:37:29 <_ikke_> _mjones: The hardware had also problems 2015-09-11 19:37:46 <_mjones> Yes, hardware does that. But I love it still. 2015-09-11 19:37:51 <_ikke_> hehe :P 2015-09-11 19:38:01 <_ikke_> But no ultimate clue 2015-09-11 19:38:37 <_ikke_> This system has been running for quite some time, but started as a VM, but migrated to run on hardware 2015-09-11 19:39:23 <_mjones> V2P is quite the unusual case. :) 2015-09-11 19:41:16 <_ikke_> Yes, it is, but the departments running this system prefer bare metal 2015-09-11 19:43:11 <_ikke_> someone here walked me through the process 2015-09-11 19:44:57 <_mjones> I'm glad someone still likes metal. 2015-09-11 19:45:33 <_mjones> Virtualisation is nigh on a miracle, but when applied ubiquitously it is easily misused. 2015-09-11 19:46:27 <_ikke_> Fun thing is that we have other departments here doing everything virtual 2015-09-11 19:47:44 is there an issue with a library known if I am getting the following error Failed to load cert/key pair: crypto/tls: failed to parse private key 2015-09-11 19:48:10 <_mjones> jmonreal: I'm not aware of any known problem with that symptom. 2015-09-11 19:48:19 _mjones: thank you 2015-09-11 19:48:30 <_mjones> Pretty obviously ASN.1 parsing. OpenSSL. 2015-09-11 19:49:18 <_mjones> Keys can take several forms. You're probably most familiar with PEM, which is the 7-bit ASCII version. DER is the native binary format. 2015-09-11 19:49:39 <_mjones> ASN.1 parsing is the bane of devs working with this. 2015-09-11 19:50:01 <_mjones> But don't jump to conclusions based on my rambling. 2015-09-11 19:51:39 _mjones: thanks 2015-09-11 19:51:45 I am using pem 2015-09-11 19:51:56 but do not know anything on how it was generated 2015-09-11 19:52:20 <_mjones> What is happening to output that error? Something TLS obviously. 2015-09-11 19:52:22 is there a restriction on the cert files on how they should be generated? 2015-09-11 19:52:23 <_ikke_> jmonreal: can you use openssl x509 to parse it? 2015-09-11 19:52:27 <_mjones> Oh, it's not apk is it? 2015-09-11 19:52:50 no, it's a software I am running on alpine 2015-09-11 19:53:32 <_mjones> Did you compile it on alpine? I think there's a soname increment between openssl 1.0.something and 1.0.{1 or 2} 2015-09-11 19:54:36 <_mjones> I don't know that pax could cause anything like this, but keep it in mind anyway. 2015-09-11 19:54:56 I do not know if you are familiar with consul 2015-09-11 19:55:56 <_mjones> I am. Haven't yet run it, though. 2015-09-11 19:56:20 <_mjones> There wasn't a package in aports yet, but I've not been looking at that in weeks. 2015-09-11 19:57:03 <_mjones> If it was me, I'd run strace on it first. 2015-09-11 20:03:53 ok 2015-09-11 20:04:04 thanks 2015-09-12 06:35:00 uh what sorts of things could go Very Wrong if I iptables -A POSTROUTING -j MASQUERADE 2015-09-12 11:05:39 nice article, http://rusty.ozlabs.org/?p=393 , even has some test to download 2015-09-12 12:48:20 vkris: that's great, thanks 2015-09-12 12:49:37 i remember tuning for packet alignment back in the dsl over atm time when it was still slow, then i stopped doing anything and just sometimes wondered "how can this be so slow" and those graphs pretty much slow its "because you didn't care any more" 2015-09-12 14:07:57 Which package holds DNS utils like `host`? 2015-09-12 14:11:27 <_ikke_> bind-tools 2015-09-12 14:12:00 _ikke_: thank you 2015-09-12 16:02:44 Hi! Is there some magical command to get build dependencies for a package? 2015-09-12 16:02:51 I'm terrible at finding stuff in wiki. :( 2015-09-12 16:04:46 abuild deps? 2015-09-12 16:07:43 Thx, found it already. 2015-09-12 16:08:00 Also on arch there were multiple helpers like yaourt to get PKGBUILD files, etc. 2015-09-12 16:08:11 Are there tools like that for alpine? 2015-09-12 17:12:37 is syslog not installed and running by default on alpine? 2015-09-12 17:12:56 i can't find a config file in /etc/syslogd.conf or see it when I run rc-status 2015-09-12 18:18:44 am i correct that there are not any man pages for apk (tried apk add apk-doc)? 2015-09-12 18:19:35 how come there's a postfix 3.0.2 here https://pkgs.alpinelinux.org/package/main/x86_64/postfix but when I run apk update all I see is 3.0.1? 2015-09-12 18:19:48 veXx: no docs by default 2015-09-12 18:22:03 no i have man installed and then "apk info | while read LINE; do apk add $LINE-doc; done" to install docs for everything installed. 2015-09-12 18:22:29 theres a quite a few that dont have man pages, i was just hoping apk did 2015-09-12 18:22:53 its easy to type man apk than to go google it 2015-09-12 18:32:47 veXx: just apk add docs and all future installs will add *-docs as well 2015-09-12 18:33:07 and there are no man pages, apk -h / apk -h is all there is atm 2015-09-12 18:33:45 doh... should have thought to do -h 2015-09-12 18:33:51 sorry, just woke up 2015-09-12 18:34:09 me doing things the hard way... 2015-09-12 18:34:19 thanks 2015-09-12 18:59:32 Am I correct in assuming that eudev-dev replaced udev-dev? 2015-09-12 19:00:30 hydrajump: postfix 3.0.2 is in the edge/main repo 2015-09-12 19:01:11 If so is it a drop-in replacement? 2015-09-12 19:09:55 clandmeter: this also affects your testing/pulseaudio, which depends on udev-dev 2015-09-12 19:15:22 veXx: ah thank you 2015-09-12 19:31:09 why does `rc-service syslog status` show it is started, but it doesn't show up in `rc-status`? 2015-09-12 19:31:15 is that normal? 2015-09-12 19:48:23 creating /etc/syslog.conf and restarting syslog doesn't seem to work as everything is still written to /var/log/messages 2015-09-12 20:15:11 ACTION is off again cya 2015-09-12 20:46:54 Do I need to first boot a CD, to create a bootablo usb with alpine? 2015-09-12 21:01:14 henriqueleng: i found this today: 2015-09-12 21:01:35 http://it-offshore.co.uk/linux/alpine-linux/48-alpine-linux-usb-hybrid-iso-image 2015-09-12 21:01:59 it's from someone here, idk who :) 2015-09-12 21:04:02 I can only get isos from the site 2015-09-13 03:36:39 ncopa, re: the comment in http://git.alpinelinux.org/cgit/aports/tree/main/qemu/musl-F_SHLCK-and-F_EXLCK.patch i'm not sure it's right 2015-09-13 03:37:02 afaik these fcntl options are for an ANCIENT legacy implementation of bsd flock() via fcntl that was used in linux 1.2 or 2.0 or something 2015-09-13 03:37:13 they should probably not be exposed to applications 2015-09-13 03:45:54 <_mjones> Is anyone awake who has insight into the default kernel config policy? 2015-09-13 03:47:07 <_mjones> I figure there's probably not a written policy for Alpine, but perhaps there's a boilerplate that distros generally follow, or perhaps Alpine is largely mimicking selections of another distro. 2015-09-13 04:00:14 <_mjones> dalias: saw your tweet about Scaleways and decided that was too irresistable to pass, so I'm planning on signing up. Thanks. 2015-09-13 04:00:55 btw i think they still have a free trial month 2015-09-13 06:12:52 <_mjones> ah, standardisation: https://en.wikipedia.org/wiki/International_Blinking_Pattern_Interpretation 2015-09-13 10:10:16 anyone around who can help me understand why the /etc/syslog.conf isn't working on alpine? 2015-09-13 13:35:31 what's the lightest weight httpd in alpine for serving static media? 2015-09-13 13:44:05 <_ikke_> perhaps lighttpd or minihttpd 2015-09-13 13:52:58 i need something with digest auth... guess that has to be lighttpd 2015-09-13 13:54:49 Trying to install py-requests with apk add py-requests on alpine 3.2 2015-09-13 13:54:59 But get this error: ERROR: unsatisfiable constraints: 2015-09-13 13:54:59 py-requests (missing): 2015-09-13 13:54:59 required by: world[py-requests] 2015-09-13 13:58:38 What do it mean? 2015-09-13 14:04:44 pnutzh4x0r: It seems like py-rquests wasn't moved to main, There is no .apk file in the main repository. 2015-09-13 14:14:40 ncopa: 2015-09-13 14:16:41 damn, ffmpeg pulls in a lot of deps 2015-09-13 14:44:21 wCPO_: it is moved in edge, just not in the stable repo 2015-09-13 14:47:31 Is there a statement from the alpine devs regarding the change of distribution policy by the grsec project? 2015-09-13 14:47:53 pnutzh4x0r: Oh, sorry for the "outdated" report then :/. Will it move to stable? or first when 3.3 get released? 2015-09-13 14:49:05 <_ikke_> fireglow: they're working on it 2015-09-13 14:49:18 <_ikke_> fireglow: trying to get in contact with the author 2015-09-13 14:51:38 _ikke_: I see, okay 2015-09-13 14:51:40 wCPO_: when 3.3 is rrle 2015-09-13 14:51:46 Released* 2015-09-13 14:52:00 Which should be next month 2015-09-13 14:52:25 U can always use repo pinning to get it now 2015-09-13 14:56:06 http://wiki.alpinelinux.org/wiki/Alpine_Linux_package_management#Repository_pinning 2015-09-13 15:01:33 argh, /usr/lib/lighttpd/mod_auth.so is not in alpine? 2015-09-13 15:02:42 oh it's a seperate package: lighttpd-mod_auth 2015-09-13 15:06:08 what's the lightest weight httpd in alpine for serving static media? 2015-09-13 15:06:11 <_ikke_> perhaps lighttpd or minihttpd 2015-09-13 15:06:19 lighthttpd is the second-most bloated, after apache 2015-09-13 15:06:25 thttpd is by far the lightest 2015-09-13 15:12:29 maybe try hiawatha 2015-09-13 15:21:58 dalias: i need digest auth 2015-09-13 15:47:20 _ikke_: hi any chance you can help me out with syslog? 2015-09-13 15:57:12 hmm 2015-09-13 15:57:45 what use is digest auth? 2015-09-13 15:59:08 if you're serving information that needs to be access controlled you really need https 2015-09-13 16:00:17 <_ikke_> hydrajump: I know too little of syslog to be of any help 2015-09-13 16:04:44 anyway i don't know much about which light httpds support auth features 2015-09-13 16:05:08 _ikke_: no worries. Thought I'd try ;) 2015-09-13 16:05:11 nginx might be your best bet 2015-09-13 16:18:16 hendry: http://www.acme.com/software/mini_httpd 2015-09-13 16:26:20 hydrajump: what about syslog? 2015-09-13 16:27:08 Sleaker: I've installed postfix and it logs to syslog by default. Out of the box everything is being logged to /var/log/messages. 2015-09-13 16:27:20 uggedal: don't think it does digest auth 2015-09-13 16:27:31 I googled and tried creating /etc/syslog.conf but no matter what I do everything is still logged to /var/log/messages 2015-09-13 16:27:34 settled on lighttpd https://github.com/kaihendry/camftp2web/blob/master/Dockerfile 2015-09-13 16:27:45 hydrajump: are you running the default install or did you actually install rsyslog or another system logger? 2015-09-13 16:28:07 if you installed rsyslog then it would already have the conf 2015-09-13 16:28:09 Sleaker: default install in terms of alpine. I have not installed any _syslog_ package 2015-09-13 16:28:14 that's your problem 2015-09-13 16:28:27 hendry: just use basic auth and tls 2015-09-13 16:28:34 I don't think you can configure the built-in syslog at all 2015-09-13 16:28:41 what I don't understand is why it's not working if there is a `syslog` service running which I have not created 2015-09-13 16:28:56 do I need to disable this syslog service before I use another? 2015-09-13 16:28:59 busybox syslog is not advanced like normal rsyslog 2015-09-13 16:28:59 <_mjones> hendry: lightest in ram usage, or what? 2015-09-13 16:29:09 the install for the others will automatically do that 2015-09-13 16:29:32 ok that wasn't intuitive to understand that I actually needed to install a syslog package 2015-09-13 16:29:41 you don't... 2015-09-13 16:30:07 <_mjones> hydrajump: it's not very obvious that default syslog is also part of busybox unless one thinks to explicitly check. 2015-09-13 16:31:20 hydrajump: the busybox syslogd can be configured like this: http://git.busybox.net/busybox/tree/docs/syslog.conf.txt 2015-09-13 16:31:40 i was confused why the syslog service was running but not showing in rc-status but perhaps that's because it's part of busybox? 2015-09-13 16:32:18 uggedal: i don't have a cert or can be bothered to set one up 2015-09-13 16:32:32 _mjones: lightest to get digest auth working 2015-09-13 16:33:03 hendry: it's most likely started due to another of your daemons having a 'need logger' 2015-09-13 16:33:08 uggedal: I tried to create that config with `mail.none -/var/log/messages` and it still logged all the postfix msgs there. 2015-09-13 16:33:41 So basically to use syslog you don't mess with the built-in busybox one, but install a package. Is this correct? 2015-09-13 16:33:43 support for /etc/syslog.conf wasn't added until about a month ago: http://git.alpinelinux.org/cgit/aports/commit/?id=6b80c507b826e38b493422719de9ef55afd4cf84 2015-09-13 16:33:57 so it's in edge, but not in stable 2015-09-13 16:34:05 hydrajump: yes 2015-09-13 16:34:19 hydrajump: no, the built in works just fine 2015-09-13 16:34:33 uggedal: no it doesn't 2015-09-13 16:34:33 if you don't have advanced needs, like remote logging etc 2015-09-13 16:34:47 Sleaker: so what don't work then? 2015-09-13 16:34:59 configuring where messages go 2015-09-13 16:36:59 that was recently added in the commit above 2015-09-13 16:37:09 exactly so it's not actually in the base install 2015-09-13 16:39:06 right, so the current solution is to use another syslog, use busybox from edge, or recompile busybox w/ that config set 2015-09-13 16:41:29 Sleaker: it works perfectly fine 2015-09-13 16:42:51 Sleaker: http://sprunge.us/WcdS 2015-09-13 16:43:48 ah, did not notice the backlog 2015-09-13 16:47:55 apparently busybox has remote logging support to if enabled. 2015-09-13 16:48:08 too* 2015-09-13 16:50:10 uggedal: was that from stable 2015-09-13 16:50:49 hydrajump: nope, edge 2015-09-13 16:51:08 ok so `apk add rsyslog` it is for now 2015-09-13 16:51:25 could backport the change 2015-09-13 16:51:53 hydrajump: you on 3.2? 2015-09-13 16:52:59 yes 3.2.2 2015-09-13 16:53:03 opps 3.2.3 2015-09-13 17:23:58 I use remote logging with stable for a while... just add SYSLOGD_OPTS="-R my.log.server" in /etc/conf.d/syslog 2015-09-13 17:24:27 and I have a syslog-ng on my.log.server 2015-09-13 18:24:27 I install rsyslog but it fails to start 2015-09-13 18:24:52 "daemon.err /etc/init.d/cron[1893]: ERROR: cannot start cron as rsyslog would not start" 2015-09-13 18:26:31 I've not touched the rsyslogd.conf config at all 2015-09-13 19:13:20 Sleaker: do you have any idea why it's failing to start? 2015-09-14 03:49:38 i've download the x86_64 iso for 3.2.3 and burned it to a USB (twice now) 2015-09-14 03:49:55 each time the target computer can't boot from it 2015-09-14 03:50:23 the sha256 matches the download: `alpine-3.2.3-x86_64.iso: OK` 2015-09-14 03:51:09 terinjokes: what do u mean burn it to usb? 2015-09-14 03:51:29 meaning i used dd to write out the disk image 2015-09-14 03:51:35 anyway, u can't just dd the iso to usb... it's not a hybrid iso 2015-09-14 03:51:42 yah that wont work 2015-09-14 03:51:43 it sux and should be fixed 2015-09-14 03:51:52 gotta use something like unetbootin or rufus. 2015-09-14 03:51:55 iirc 2015-09-14 03:51:57 ah, i've just done it the same way i've always done it 2015-09-14 03:52:08 I use rufus. 2015-09-14 03:52:20 but that's for windows 2015-09-14 03:52:45 i remember rufus 2015-09-14 03:52:45 http://it-offshore.co.uk/linux/alpine-linux/48-alpine-linux-usb-hybrid-iso-image 2015-09-14 03:53:04 http://wiki.alpinelinux.org/wiki/Create_a_Bootable_USB 2015-09-14 03:54:13 it's actually pretty easy 2015-09-14 04:03:27 used unetbootin and still had no luck 2015-09-14 04:05:10 use the link sleaker gave 2015-09-14 04:05:22 it was real easy for me to follow and im a noob 2015-09-14 04:11:33 something must be weird with alpine, it's not recognized by my machine as bootable 2015-09-14 04:11:46 even with the hybrid iso image 2015-09-14 04:12:53 if i dd ubuntu onto the same stick, the live environment boots fine 2015-09-14 04:16:31 <_mjones> terinjokes: obvious question: UEFI or legacy BIOS? UEFI support isn't in the current images, but could make it into 3.3.0. 2015-09-14 04:17:08 atleast its supports non-pae systems lol 2015-09-14 04:17:13 UEFI is disabled 2015-09-14 04:17:42 my little router cant run ubuntu server 2015-09-14 04:23:25 i just checked the BIOS and double checked that UEFI is disabled 2015-09-14 04:26:46 veXx: oh the base grsec image has PAE disabled? 2015-09-14 04:26:48 odd. 2015-09-14 04:27:00 PAE is like built-in to the kernel since 3.x 2015-09-14 04:28:24 <_mjones> Do you know the kernel config variable? 2015-09-14 04:28:53 <_mjones> The closest I can find is CONFIG_PHYS_ADDR_T_64BIT, but PAE isn't 64-bit, so 2015-09-14 04:29:37 <_mjones> oh well, bing to the rescue! 2015-09-14 04:29:56 If building 32 bit kernel make sure you have CONFIG_X86_PAE enabled (which is set by selecting CONFIG_HIGHMEM64G) 2015-09-14 04:30:11 funny, story i got this astaro appliance from a friend but its old, non-pae old... installed alpine 32-bit no problem 2015-09-14 04:30:31 non-PAE doesn't necessarily mean old 2015-09-14 04:30:49 VIA makes 586 non-PAE CPUs 2015-09-14 04:31:22 <_mjones> I wish via made modern CPUs. 2015-09-14 04:31:38 via c3 2015-09-14 04:32:00 :D 2015-09-14 04:32:21 <_mjones> I don't know that any of the boards I've been looking at have vias, and I've been looking as low as amd t40e bobcat (pcengines apu) and intel z3735f (intel compute stick; various terribly cheap chinese tablets) 2015-09-14 04:32:58 honestly the vias are pretty crappy 2015-09-14 04:33:01 <_mjones> If I have to use x86 architecture, I want a lot more choices. 2015-09-14 04:33:02 way overpriced 2015-09-14 04:33:09 you might as well get the mini i3s 2015-09-14 04:33:42 it got 2 astaro security gateway for free... as in beer 2015-09-14 04:33:44 <_mjones> They used to sell in decent numbers. Now I don't know what their market is. 2015-09-14 04:34:54 im to much of a noob with linux for alpine, so im installing debian... it sucks cause i really liked that alpine has grsec built in 2015-09-14 04:35:14 but just dont have support for some of the packages im looking for... yet 2015-09-14 04:35:17 ^_^ 2015-09-14 04:35:24 which ones vexx? 2015-09-14 04:35:31 aide for starters 2015-09-14 04:36:21 i did see you guys have squid 2015-09-14 04:36:47 i really hope alpine picks up 2015-09-14 04:37:37 i did consider trying to package stuff myself to help out... but like i said im a noob 2015-09-14 04:37:58 centos and ubuntu are what i know 2015-09-14 04:38:30 Sleaker, _mjones: should i try the overlay method from the ubuntu live environment? 2015-09-14 04:46:38 <_mjones> terinjokes: I think I only vaguely know what you mean, overlay method. 2015-09-14 04:46:46 http://wiki.alpinelinux.org/wiki/Replacing_non-Alpine_Linux_with_Alpine_remotely 2015-09-14 04:47:00 <_mjones> VeXx: advanced intrusion detection environment? I might look into doing a build for that. 2015-09-14 04:47:24 <_mjones> terinjokes: oh. I only vaguely remember that article. 2015-09-14 04:47:51 <_mjones> I could actually use that right now. Well, I'd have to compile a tree for armel instead of armhf, but still. 2015-09-14 04:49:02 _mjones: that would be the one... not having it on a system im less familiar with makes me nervous 2015-09-14 04:49:14 <_mjones> terinjokes: to be honest I don't think I would in your case. If you could figure out your boot issue, you might be installing in two minutes. 2015-09-14 04:49:59 yeah, certainly easier if I can boot normally 2015-09-14 04:50:17 <_mjones> veXx: I'm making a note into checking into contributing a package build. No promises, of course. Anything else on your wishlist? 2015-09-14 04:50:55 <_mjones> I'm pretty far behind in cleaning up and contributing the package builds I have made. So...slack. 2015-09-14 04:51:51 <_mjones> Sometimes they're nearly trivially easy. Other times, not so much. So far I've found the big c++ packages to be the most troublesome, especially if they're less mature and frequently changing. 2015-09-14 04:53:07 <_mjones> speaking of squid....I actually can't remember if I tried trafficserver on alpine, even while I was working with trafficserver code. I really should have done that. 2015-09-14 04:53:23 <_mjones> actually trafficserver is ++. Let me check my devbox. 2015-09-14 04:54:31 bind-chroot might be nice, but really not required as it can be manually done 2015-09-14 04:55:18 <_mjones> Yes, looks like trafficserver was a bit of a bother and I didn't continue working on it. Stopped on 2015-06-21, looks like. 2015-09-14 04:56:11 <_mjones> Some of the ones I did were already contributed by someone. Good for them. I really need to get set up for that. 2015-09-14 04:56:14 i was looking to use it as a light weight firewall/router 2015-09-14 04:56:27 i cant think of anything else atm 2015-09-14 04:57:01 <_mjones> It's really very much a general-purpose linux. Quite a few people running on desktops and laptops. 2015-09-14 04:57:58 yea i looked through the packages, i was surprised 2015-09-14 04:58:10 <_mjones> I was going to Alpine my new desktop but had uefi problems and went with another distro for a moment. Probably a good decision until I know exactly the limits of this hardware (important in my use-case). 2015-09-14 04:58:29 <_mjones> OK: 8448 distinct packages available 2015-09-14 04:59:06 <_mjones> veXx: that's not into debian territory with 30,000 or 40,000, but it's a hell of a lot more than I used to have on rhel/cent years ago. 2015-09-14 04:59:29 Query: does Alpine have support for ZFS On Linux? The idea is to use it as a minimalist file server (no routing). 2015-09-14 04:59:37 <_mjones> They tend to be quite up to date, although not quite as aggressively as Arch. 2015-09-14 04:59:59 <_mjones> Let me see if there's a zfs kernel. 2015-09-14 05:01:10 <_mjones> Ah. There's a zfs-grsec kernel build in testing, so that seems to be a tentative 'Yes'. 2015-09-14 05:01:27 _mjones: yea uefi... lets not go there lol 2015-09-14 05:01:50 <_mjones> You'd have to enable testing repo. Consensus is that testing repo is best employed if you're running 'Edge' version of alpine instead of stable release. 2015-09-14 05:02:54 <_mjones> uefi is flawed, but all art is flawed. the principles were fairly sound, but among other things the standard was too far influenced by microsoft. I'm not a uefi expert, though. 2015-09-14 05:03:25 <_mjones> bios is certainly a hideous legacy mess. bios is one of the reasons I wouldn't run PCs for many years. 2015-09-14 05:03:59 Ah. 2015-09-14 05:04:17 So Alpine would not be recommended for production quality NAS systems. Thanks though! 2015-09-14 05:04:46 <_mjones> btw, you can look at the complete aports tree on http://git.alpinelinux.org/ . So you can easily look for packages without installing Alpine. 2015-09-14 05:04:57 _mjones: oh, that was the other thing, i was supprise that you didnt have MAC like apparmor or selinux since you guy grsec... but im sure thats one of those starwars dreams... off in a distant galaxy far far away lol 2015-09-14 05:05:29 <_mjones> sachiru: I have nothing but respect for zfs, but many people will be as well served by xfs or btrfs, I think. 2015-09-14 05:06:10 _mjones: yea microsoft makes me weep 2015-09-14 05:06:19 <_mjones> VeXx: I'm actually not at all sure about the compatibility of linux security modules with grsec; does grsec count as a security module? Been meaning to look into it. 2015-09-14 05:07:48 <_mjones> sachiru: facebook is running btrfs on a subset of production systems, and is aggressively pushing forward with it. Lead btrfs dev Chris Mason works for facebook. 2015-09-14 05:08:02 I personally prefer ZFS as a NAS system. Linux on a RAID6 BTRFS? It becomes Break-The-Root-FS. 2015-09-14 05:08:39 The statement about Alpine as root refers to Alpine support for ZFS, not alpine or zfs in general./ 2015-09-14 05:08:49 <_mjones> Sachiru: can you further explain the failure mode(s) that concern you? 2015-09-14 05:08:57 Alpine? Very good. ZFS? Very good. Glue between Alpine and ZFS? Not so very good. 2015-09-14 05:09:17 <_mjones> Sachiru: no, I understand and that's not a problem. 2015-09-14 05:09:44 <_mjones> Sachiru: I wouldn't run it production either. But I wouldn't run zfs production on linux at all, fuse or not. 2015-09-14 05:10:17 <_mjones> If I really wanted zfs I'd have to run freebsd or one of the illumoses. I guess OmniOS. 2015-09-14 05:10:36 _mjones: not sure exactly what your asking. but grsec patches makes exploitation harder (aslr and all that jazz) 2015-09-14 05:11:10 <_mjones> raid5 and raid6 code was only contributed to btrfs comparatively recently. Compare with about a decade of maturity for zfs as a whole. 2015-09-14 05:11:37 <_mjones> vexx: that's actually PaX for aslr, etc. 2015-09-14 05:11:47 I've personally run ZFS on Ubuntu/Debian for production systems. So far it's been very good. 2015-09-14 05:12:15 However, Ubuntu is a bit more heavy than Alpine, and I'm looking for lightweight alternatives that can work as NAS boxes and *only* NAS boxes. 2015-09-14 05:12:20 _mjones: yes, pax is the part of grsecurity that adds aslr 2015-09-14 05:12:26 <_mjones> VeXx: I don't know about compatibility between grsec and (selinux|apparmour|smack|tomoyo). 2015-09-14 05:12:57 <_mjones> sachiru: kernel zfs or fuse? 2015-09-14 05:14:08 Kernel module. 2015-09-14 05:14:13 Fuse is a bit too slow. 2015-09-14 05:14:28 <_mjones> yes, one assumes. 2015-09-14 05:15:41 <_mjones> I used to run some production heavy nfs nas over 10gbase and infiniband. Performance is important to me. 2015-09-14 05:16:11 _mjone: oh yea that reminds me, since the big announcement about grsec what are you guys going to do? use their test series? 2015-09-14 05:17:36 <_mjones> veXx: ncopa is in contact with grsecurity team. 2015-09-14 05:18:35 nice 2015-09-14 05:18:52 <_mjones> VeXx: it looks like things were going to be resolved in some favourable way, as of the end of last week. 2015-09-14 05:19:47 <_mjones> VeXx: according to ncopa, it would be possible to continue with test series but would be a significantly larger maintenance effort. 2015-09-14 05:20:20 _mjones: i found you guys through distrowatch under firewalls, you were like #2 and as soon as i saw grsec... i was sold 2015-09-14 05:21:51 <_mjones> a number of people have asked about the grsecurity situation, especially me. 2015-09-14 05:21:53 _mjones: i really been wanting to learn hardened gentoo (grsec, selinux or apparmor) but havent had the time 2015-09-14 05:22:18 <_mjones> Maybe a news item about it would be a good idea. 2015-09-14 05:22:39 hell, just drop something in the topic 2015-09-14 05:22:47 lol 2015-09-14 05:22:53 <_mjones> also a good idea. ;) 2015-09-14 05:23:04 <_mjones> I'm not +o. 2015-09-14 05:30:30 <_mjones> damn, my ublock extension is crashing and preventing me from accessing sourceforce. VeXx, could you possibly send me a full url for the latest AIDE source tarball? 2015-09-14 05:31:19 give me a sec 2015-09-14 05:32:56 http://downloads.sourceforge.net/project/aide/aide/0.15.1/aide-0.15.1.tar.gz?r=http%3A%2F%2Fsourceforge.net%2Fprojects%2Faide%2Ffiles%2F&ts=1442208729&use_mirror=skylineservers 2015-09-14 05:34:18 <_mjones> I'm not doing an apkbuild just yet, I'm giving it a preliminary look to see if it's pretty trivial or not. 2015-09-14 05:34:24 let me know if that doesnt work 2015-09-14 05:34:58 <_mjones> the thing is musl. Musl is superb, but because it's not glibc and its headers are not glibc, things sometimes need some massaging. 2015-09-14 05:35:26 yea i read on the wiki about that... which is why i didnt even attempt lol 2015-09-14 05:35:35 <-- straight newb 2015-09-14 05:36:01 <_mjones> die gnu autotools. Let's see how it goes. 2015-09-14 05:36:26 its why i didnt run linux for a long time... i can lock a windows box down easy, linux terrified me 2015-09-14 05:36:42 but i grew up with windows... unfortunately 2015-09-14 05:38:54 <_mjones> I claim no windows expertise. But 'lock a windows box down' sounds rather dubious to me. At least if it still functions afterwards. 2015-09-14 05:39:05 <_mjones> Very difficult to turn down services arbitrarily. 2015-09-14 05:39:08 lol 2015-09-14 05:39:11 <_mjones> like workstation. 2015-09-14 05:39:41 <_mjones> Linux is in many ways a lot simpler. 2015-09-14 05:40:03 <_mjones> fewer lines of code. Yet equally as capable, if not more so. 2015-09-14 05:40:11 for work stuck, i agree... for personal computer, i ran comodo and customized the crap out of it 2015-09-14 05:40:26 your preaching to the choir 2015-09-14 05:40:38 im done with windows 2015-09-14 05:41:13 but want to have the same level of understanding (or greater) than i did of windows 2015-09-14 05:41:40 <_mjones> That's a very reasonable requirement. 2015-09-14 05:42:25 <_mjones> aide ./configure is not finding my libgpg-error. Is it possible there's no .a installed? Hmm. 2015-09-14 05:42:29 kind disheartens me that windows has EMET but most distros dont use grsec 2015-09-14 05:43:20 actually it would terrify me if i didnt run noscript lol 2015-09-14 05:43:57 <_mjones> noscript is needy. These days I'm on privacy badger and ublock origin, basically. 2015-09-14 05:44:11 ive heard a lot of good things about ublock 2015-09-14 05:44:25 <_mjones> Just don't run flash, java plugin, or adobe pdf reader and run a current browser and you're fine. ;) 2015-09-14 05:44:34 but frankly, i dont mind doing the leg work for security 2015-09-14 05:44:58 hell id be on hardened gentoo, if i had the extra time for learning curve atm 2015-09-14 05:45:08 <_mjones> grsec isn't part of mainline kernel, like zfs isn't and xen wasn't (isn't?) 2015-09-14 05:45:13 but im trying to learn assembly atm 2015-09-14 05:45:43 <_mjones> that's another place the posix world and the dec/dos/win world are entirely different. 2015-09-14 05:45:52 yep 2015-09-14 05:46:01 <_mjones> windows asm uses intel convention for format. unix asm uses at&t convention. 2015-09-14 05:46:15 yea, im learning the at&t syntax 2015-09-14 05:47:15 <_mjones> so, I'm pausing on aide for now. It wants the .a library, which is the library for compiling static. That isn't installed on Alpine for libgpg-error. 2015-09-14 05:47:23 i spent wrote 100k lines of code and i got was hello world 2015-09-14 05:47:24 lol 2015-09-14 05:47:46 thats awsome 2015-09-14 05:48:22 <_mjones> This hasn't come up before, and I'm now curious about the policy over whether to include them or not. Obviously, some/many libs get .a installed. 2015-09-14 05:48:59 <_mjones> It might be that everything should have a .a, and the aports tree needs to be updated to reflect that. Or it may be that policy says something else. I have no idea. 2015-09-14 05:49:23 over my head 2015-09-14 05:49:54 <_mjones> algitbot needs to keyword spew a few more XKCDs I think. 2015-09-14 05:51:10 <_mjones> VeXx: compiling static means all dependency libraries are included in the one executable; dynamic linking means they're linked at runtime. .a libraries are static, .so libraries are dynamic. 2015-09-14 05:51:19 <_mjones> haha. 2015-09-14 05:52:03 <_mjones> Things are typically dynamic on normal dynamic systems, but a few things can always be done static for disaster recovery, security, performance, or maintainability if desired. 2015-09-14 05:52:11 ahh 2015-09-14 05:52:41 <_mjones> Use the command 'file' on something and you'll get the hang of it. ;) 2015-09-14 05:53:40 <_mjones> Type this to see what I mean: file mount /lib/libc* 2015-09-14 05:53:53 <_mjones> file is so choice. 2015-09-14 05:54:08 usr/bin/aide: ELF 64-bit LSB executable, x86-64, version 1 (SYSV), statically linked 2015-09-14 05:54:13 <_mjones> I've not seen its like on MS systems, which is both surprising and a shame. 2015-09-14 05:54:14 yep, statically linked 2015-09-14 05:54:15 lol 2015-09-14 05:55:11 yea, microsoft prefers dll hell 2015-09-14 05:55:14 <_mjones> file uses magic numbers -- some bytes at the start of a file -- to distinguish the types. 2015-09-14 05:55:38 yea i havent learned about elf layout yet 2015-09-14 05:55:47 but ill get there soon enough 2015-09-14 05:56:34 yea, microsofts solution for all the dynamic library conflicts was winsxs 2015-09-14 05:56:35 <_mjones> how many years were you on windows? 2015-09-14 05:56:41 you know what that is? 2015-09-14 05:56:47 20 years 2015-09-14 05:56:50 <_mjones> somewhat, yes. 2015-09-14 05:57:13 <_mjones> twenty. Back to 3.1, then. 2015-09-14 05:57:23 its a tragedy and means after a few years, your C:\windows directly is 100GB 2015-09-14 05:57:25 lol 2015-09-14 05:57:30 stop telling my age 2015-09-14 05:57:38 and yes 2015-09-14 05:57:56 <_mjones> winsxs is something like little virt environments, or little chroots, to separate out dll environments. Something like that, right? 2015-09-14 05:58:06 yea 2015-09-14 05:58:50 so now ask yourself, when you update a library in windows for security purposes... and its not really removed... what bad things can happen... 2015-09-14 05:59:11 but they tell you not to touch that directory 2015-09-14 05:59:19 <_mjones> I started on unix in 1993. Never did much with windows. Touched it some, have admined more than a few, but I never actually ran windows for anything so obviously I never had to convert. 2015-09-14 05:59:44 <_mjones> Interesting. Java runtimes also don't remove the old one by default. 2015-09-14 05:59:47 i didnt have any mentors unfortunatly 2015-09-14 05:59:59 <_mjones> A real smh moment when I learnt that. 2015-09-14 06:00:00 so just bad choice one after another 2015-09-14 06:00:15 so when i learned to program... it was basic -> visual basic 2015-09-14 06:00:25 but ive purged that from my system 2015-09-14 06:00:29 <_mjones> yeah, that. 2015-09-14 06:00:43 decide to start with a real language.. assembly 2015-09-14 06:00:44 lol 2015-09-14 06:00:52 <_mjones> I'm of the opinion now that few things have stunted brains and hurt CS more than basic. 2015-09-14 06:01:44 after, ill probably learn python so i can do stuff quickly, then loop back around to C 2015-09-14 06:01:48 <_mjones> I had some too. Perversely enough, I was briefly responsible for a (niche, vertical market) ERP system run entirely in interpreted basic. Running on unix. 2015-09-14 06:01:54 but im not a programmer, im a security guy 2015-09-14 06:02:11 lol 2015-09-14 06:02:12 <_mjones> everyone is a programmer. I'm just a bad one. 2015-09-14 06:02:12 wtf 2015-09-14 06:02:27 well some people literally don't program 2015-09-14 06:02:36 they do lkadsfjkl;adsfjkl;adfsjkl;adfsjkl;adefsjkl;aedfs 2015-09-14 06:02:43 while adsjkladfsjkl;adsfjkl;adfsjkl; done 2015-09-14 06:02:45 <_mjones> software for verticals will terrify you like an h.p. lovecraft short story. Truly. 2015-09-14 06:03:00 software for verticals? 2015-09-14 06:03:05 verticals? 2015-09-14 06:03:30 <_mjones> Yes, there are people who are paid to program who to all observation seem worse than I. 2015-09-14 06:03:32 https://en.wikipedia.org/wiki/Vertical_market this sort? 2015-09-14 06:03:42 <_mjones> vertical markets. It's an industry term. 2015-09-14 06:03:52 <_mjones> So one example would be professional offices. 2015-09-14 06:04:18 <_mjones> Doctors' offices have special doctor's office software, lawyers' offices have special lawyer office software. 2015-09-14 06:04:31 ohh 2015-09-14 06:04:33 the kind where there's basically no competition because there isn't anyone better who's going into the market? 2015-09-14 06:04:38 <_mjones> The market is specialised and so the scale on which the software can be sold is limited. 2015-09-14 06:04:48 yea i have nightmares about that stuff 2015-09-14 06:04:48 because it's a tiny market 2015-09-14 06:05:19 <_mjones> ammunta: yes, and no individual customer can justify the capital to have the whole mess replaced, probably not even if you started with an open-source ERP system. 2015-09-14 06:05:33 I work for one 2015-09-14 06:05:42 we do Point of Sale software for Quick-Lubes 2015-09-14 06:05:43 so you start off on a bad foot and then you just accumulate worse 2015-09-14 06:05:43 <_mjones> These systems generally consist of industry-specialised ERP systems, basically. 2015-09-14 06:05:44 yea, its the type of software that makes people say... well we have to still run windows 2000 because our software will not run on xp 2015-09-14 06:05:45 it'sstupid 2015-09-14 06:05:45 lol 2015-09-14 06:05:50 sooo crappy 2015-09-14 06:06:02 Sleaker, _mjones: even upgraded my BIOS with no cigar. officially out of ideas 2015-09-14 06:06:03 the only reason why the company is staying afloat is because they have no legitimate competition 2015-09-14 06:06:27 <_mjones> Sleaker: that's pretty typical, in my thankfully-limited experience. 2015-09-14 06:07:03 <_mjones> The basic system I mentioned. We would be issued patches...as a handful of numbered lines...to be pasted into the prod app. 2015-09-14 06:07:19 <_mjones> The....horror. The horror you cannot believe. 2015-09-14 06:08:09 terinjokes: you may try creating a VM, mount the iso and startup. then run setup-bootable and point it to your usb stick 2015-09-14 06:08:14 <_mjones> As soon as a player figures out how to genericise these requirements and starts selling a modern open-source ERP customised for them, it will be the decline of the vertical markets. 2015-09-14 06:08:30 really trying to find another job actually 2015-09-14 06:10:24 sleaker, have you considers writing hacks for games lol? i know theres 1 site that the guy maintains hacks for like 20 games, charges $20/month and has over 100k active subs 2015-09-14 06:10:34 lol no 2015-09-14 06:10:40 I'm not that good at reverse engineering 2015-09-14 06:10:43 <_mjones> omg. 2015-09-14 06:10:47 i saw that and it really made me think about learning to right cheats hahaha 2015-09-14 06:11:08 <_mjones> Those are the kind of cracks that have to be updated frequently in order to continue to work, right? 2015-09-14 06:11:24 <_mjones> I imagine he can afford a couple of licensed copies of ida pro, yeah. 2015-09-14 06:11:25 sometimes yes 2015-09-14 06:11:27 yea... but you just need to stay open for like 3 months 2015-09-14 06:11:32 then cash in and bro down 2015-09-14 06:11:53 thats more than id make in several years 2015-09-14 06:11:53 <_mjones> you mean the individual game or the whole site? 2015-09-14 06:12:20 <_mjones> I imagine it takes a long time to get to the reputation to support those kinds of numbers. 2015-09-14 06:13:06 this is my retirement plan... learn assembly, learn reverse engineering, learn c, learn the shit out of rootkits, write a few cheats and take family to my new island 2015-09-14 06:13:50 it kinda makes me sad cause im a gamer... but id be ok getting stomped on by somebody paying me $20/month 2015-09-14 06:13:59 <_mjones> I hate to say this, but all the rootkit money is in windows. ;) 2015-09-14 06:14:07 yep 2015-09-14 06:14:45 <_mjones> actually, the original meaning of rootkit was fairly different. Not that many people know or remember that. 2015-09-14 06:15:43 my theory is, you dont have to worry about updating your cheats to frequently if you can manage a really great rootkit 2015-09-14 06:15:47 :D 2015-09-14 06:16:05 if windows doesnt know about it, vac and seeing it 2015-09-14 06:16:25 <_mjones> I just realised you wrote 100k subscribers. I actually read that as 1,000 subscribers for some reason. 2015-09-14 06:17:08 well, thats what it was when i checked a year ago 2015-09-14 06:17:10 <_mjones> veXx: you ought to read up on secure boot, tpm, verified boot, and...wait, I'll grab a URL. 2015-09-14 06:17:54 <_mjones> review of highly relevant book: http://firmwaresecurity.com/2015/08/01/book-review-platform-embedded-security-technology-revealed/ 2015-09-14 06:17:57 i have 2015-09-14 06:18:36 <_mjones> There are a lot of things in that book that have no substantial coverage in the open literature online. 2015-09-14 06:19:12 all this stuff came about because the gov was hearing about bios and firmware base "implants" 2015-09-14 06:19:20 nice 2015-09-14 06:19:36 ill bookmark it for later 2015-09-14 06:19:50 <_mjones> well now you know why they were so concerned about theoretical attacks. 2015-09-14 06:20:15 ive known for years 2015-09-14 06:20:40 i do security by trade 2015-09-14 06:20:58 but ive always been paranoid 2015-09-14 06:21:00 <_mjones> This is a problem, though, because intel's efforts are about putting the hardware in a position of control over the OS, and not trusting the OS, and DRM. Because intel's customers are somewhat interested in this, and intel sells hardware and not (really) software. 2015-09-14 06:21:44 not to mention all the other "features" theyve added to their chips 2015-09-14 06:21:47 <_mjones> But putting hardware in a position of control over software is fundamentally at odds with Software Defined everything. 2015-09-14 06:22:35 <_mjones> Now we have the FCC looking to start the clampdown on something I've been surprised they've been tacitly ignoring. 2015-09-14 06:22:59 whats that? 2015-09-14 06:23:13 <_mjones> Have you seen the 'help us save wifi'? 2015-09-14 06:24:20 ohh yea i heard about them wanting to crack down on people modding their wifi stuff 2015-09-14 06:24:25 is that what your talking about? 2015-09-14 06:24:57 <_mjones> In an effort to stop the potential for users to bypass WiFi DFS (some frequencies that clash with weather radar in some jurisdictions; requires special compliance measures), FCC is on the verge of promulgating rules that would mandate measures to prevent users from flashing open-source firmware. 2015-09-14 06:25:01 <_mjones> yes, that. 2015-09-14 06:25:19 <_mjones> That seems fairly specific, and it is. 2015-09-14 06:25:47 <_mjones> But you can count on government to stick to a principle only when it's most inconvenient or insidious. 2015-09-14 06:25:59 <_mjones> So the actual thread is to all software defined radio. 2015-09-14 06:26:06 <_mjones> s/thread/threat/ 2015-09-14 06:26:22 <_mjones> And all radio will become software defined, so... 2015-09-14 06:32:11 I'm kinda proud of myself 2015-09-14 06:32:55 teaching myself C/C++ 2015-09-14 06:33:11 and learned Cmake pretty well the last couple days 2015-09-14 06:39:32 can't you compromise windows by just plugging in a USB drive? 2015-09-14 06:39:35 <_mjones> imho, c and c++ are not much alike in practise. 2015-09-14 06:39:54 _mjones: yah I'm aware. C++ is going to be much mroe like my Java if done idiomatically 2015-09-14 06:40:05 just with slightly more mem management thrown in 2015-09-14 06:40:20 <_mjones> You can compromise most things by plugging in firewire, and most or a lot by plugging in USB, depending on semantics. 2015-09-14 06:40:40 <_mjones> idiomatically is the right way to regard it, yes. 2015-09-14 06:40:57 <_mjones> And Windows idioms are dramatically different than posix. 2015-09-14 06:41:12 <_mjones> Most ++ I run into is highly informed by windows idioms. :( 2015-09-14 06:41:44 <_mjones> algitbot needs regex upgrade. 2015-09-14 06:42:10 yah. 2015-09-14 06:42:26 I develop POS software in Java/Perl/dataflex on Linux 2015-09-14 06:42:44 <_mjones> Hey, Linux. Could be worse. 2015-09-14 06:42:57 I don't get to run linux on my development machine 2015-09-14 06:43:02 I'm forced to do everything through a VM ;-/ 2015-09-14 06:43:45 <_mjones> that's an enterprise fad that is on the way out, for the most part. 2015-09-14 06:44:17 yah I dunno. I think at one point all the machines in the office were dummy terminals 2015-09-14 06:44:30 so at some point they thought it was a good idea to swap to windows boxes 2015-09-14 06:46:16 <_mjones> I'm a fan of thin client for a lot of things. Laptops put a spanner in those works some time ago. 2015-09-14 06:46:26 yah no laptops. 2015-09-14 06:46:29 desktops only 2015-09-14 06:46:57 <_mjones> Modern 'VDI' is not really the same. MS licensing costs are high with today's VDI, which is one reason MS is so happy weith it. 2015-09-14 06:47:30 <_mjones> No, the real breakthrough in thin clients I recognised when I saw. ChromiumOS/ChromeOS. 2015-09-14 06:47:39 that's not thin client though 2015-09-14 06:47:48 that's just a slimmed down OS 2015-09-14 06:48:01 <_mjones> Sure it is. The only content stored locally is cached. 2015-09-14 06:48:29 <_mjones> OK, ok. Depends on semantics. I'm concerned mostly with stateless clients, and also with security. 2015-09-14 06:48:36 yah.. hehe 2015-09-14 06:48:39 it's not all cached 2015-09-14 06:48:43 there's local docs. 2015-09-14 06:48:45 <_mjones> Where the actual cycles are consumed is subject to optimisation, not principle. 2015-09-14 06:48:47 and you can download stuff 2015-09-14 06:49:45 omg... the flying imps at my work do nothing but talking about all the promises vdi will bring us... please please please... make it stop 2015-09-14 06:49:45 <_mjones> Sleaker: yes, but....close to semantics. And certainly disableable in a ChromiumOS build. 2015-09-14 06:50:26 <_mjones> Don't worry. Right after they take it production, at massive cost, it will become legacy. ;) 2015-09-14 06:50:36 lol 2015-09-14 06:51:00 <_mjones> The crest of the VDI hype wave is already receding into the distance. 2015-09-14 06:51:12 they hire windows administrators to manage their unix systems... because they are cheaper 2015-09-14 06:51:15 hahahahahaha 2015-09-14 06:51:28 i doubt it will ever make production 2015-09-14 06:52:18 and always ask... whats the bare minimum i have to do security wise to get this blessed 2015-09-14 06:52:41 i want to say... never ask me that question again... good day... I SAID GOOD DAY SIR 2015-09-14 06:53:10 <_mjones> It will persist for a little longer for those whom fantasize about running heavy CAD workstations from cheap little laptops at home. 2015-09-14 06:53:39 <_mjones> vexx: at least they're honest when they ask what is the bare minimum. 2015-09-14 06:53:49 lol 2015-09-14 06:54:05 <_mjones> I mean you never go into a sox or pci-dss review and declare that you firmly intend to go above and beyond on all criteria. 2015-09-14 06:54:33 <_mjones> Also, my tax remittances will be above and beyond. 2015-09-14 06:57:16 <_mjones> people really hire windows admins for unix? 2015-09-14 06:57:28 Sounds scary 2015-09-14 06:57:48 Most of the Windows admins I know hardly know how to script the most simple tasks 2015-09-14 06:57:55 <_mjones> I've heard it was always one attraction of windows during the initial uptake, 1995-2005, but that was for windows machines. 2015-09-14 06:58:03 When it comes to administrating Linux, scripting skills are a must 2015-09-14 06:59:03 <_mjones> The leading-edge windows people are busy telling each other that powershell is a must, over and over. So don't worry too much. 2015-09-14 06:59:48 From what I have been reading, the graphical interface for Windows server is dropped for Server 2016. Now it is optional. 2015-09-14 06:59:58 So yes, they will have to learn powershell 2015-09-14 07:00:28 lol 2015-09-14 07:00:50 <_mjones> engblom: Server Core dates back to Server 2008. The problem is third-party stuff rarely works just as well without the gui. 2015-09-14 07:01:49 <_mjones> typical deployments got pretty frustrated with server core because not everything snaps into the remote management framework, etc. 2015-09-14 07:02:11 people look at me crosseyed when i tell them powershell some great invention thats going to modernize computing... nix has had it for decades and its called shell scripting 2015-09-14 07:02:28 they look at me like i just cursed god himself 2015-09-14 07:03:09 sorry, that powershell is NOT 2015-09-14 07:03:38 rather that its microsofts attempt to bring to it what nix has had all along 2015-09-14 07:03:53 <_mjones> lack of scriptability was always a problem on macos and windows, all the way back. 2015-09-14 07:04:19 batch scripting was a ***** 2015-09-14 07:04:45 <_mjones> This was fairly deliberate on windows, in my estimation. By confining most automation to language APIs, instead of scriptable CLI, Microsoft deliberately opened a huge opportunity for ISVs to profit. 2015-09-14 07:05:56 <_mjones> In fact, a friend of mine managed to hack together a decent IPAM system that integrated with NT4 dns and dhcp by pushing the cli api to the limit. 2015-09-14 07:06:31 <_mjones> Was production in our shop. In retrospect, I should have strongly considered taking that commercial at the time. 2015-09-14 07:07:04 <_mjones> So, my theory is that lack of scriptability was largely intentional. 2015-09-14 07:07:49 interesting theory 2015-09-14 07:12:44 <_mjones> The broader windows culture has always accepted paying for smaller software packages, going back all the way to cp/m. But nobody could sell scripts even there. 2015-09-14 07:12:52 <_mjones> developers! developers! developers! 2015-09-14 07:14:35 i think the firewall i been working on for 3 days just died... 2015-09-14 07:14:54 /facepalm 2015-09-14 07:16:33 night 2015-09-14 07:17:13 morning 2015-09-14 07:18:18 morning 2015-09-14 07:20:04 <_mjones> goodnight all. 2015-09-14 07:20:29 <_mjones> timezones. how do they work? 2015-09-14 07:20:34 :) 2015-09-14 07:57:16 morning 2015-09-14 07:57:31 firefox from testing (40.0.3) does not work 2015-09-14 08:00:00 firefox from main (38.1.1) works fine however 2015-09-14 08:11:52 about firefox: forget it, I was mixing stable and edge repositories 2015-09-14 08:48:24 ugh this syslog thing on alpine is super annoying. The only thing I'm really have trouble with so far on alpine :( 2015-09-14 08:48:34 s/have/having 2015-09-14 09:07:22 also this user is created by the postfix package `vmail:x:100:208:Linux User,,,:/var/mail/domains:/sbin/nologin` but I need to change the UID to > 500. I tried usermod but the command doesn't exist. 2015-09-14 09:08:48 hydrajump: what's syslog not playing? 2015-09-14 09:10:39 ScrumpJack: hey so I added `rsyslog` as was suggested to me last night. However when I do rc-update add rsyslog default and rc-service rsyslog start it fails to start and says 2015-09-14 09:10:41 "daemon.err /etc/init.d/cron[1893]: ERROR: cannot start cron as rsyslog would not start" 2015-09-14 09:10:55 I haven't made any changes to the rsyslog.conf 2015-09-14 09:11:46 hmm 2015-09-14 09:11:59 let me check here 2015-09-14 09:13:14 thanks 2015-09-14 09:14:24 how come cron was stopped? 2015-09-14 09:15:29 no idea I just rebooted the system and it fails to start when I have rsyslog starting at boot. As soon as I removed rsyslog `rc-update del rsyslog` and rebooted cron worked fine 2015-09-14 09:15:41 ah, ok, you rebooted. hang on 2015-09-14 09:23:41 looks like rsyslogd doesn't like the option -c5 2015-09-14 09:25:10 so something is broken witht the rsyslogd package? 2015-09-14 09:52:22 hydrajump: could you raise a bug report on bugs.alpinelinux.org? 2015-09-14 10:04:51 ok do you have any solution in the meantime? There must be some way to get syslog to work so that all the logs don't go into /var/log/messages 2015-09-14 10:05:05 I'm surprised no one has experienced this 2015-09-14 10:20:45 just comment out RSYSLOG_OPTS="-c5" in /etc/conf.d/rsyslog 2015-09-14 10:24:06 I suspect most alpine linux users are using syslog or syslog-ng 2015-09-14 10:24:31 ScrumpJack: https://bugs.alpinelinux.org/issues/4640 2015-09-14 10:25:03 oh I would not have installed rsyslog if it wasn't for the built-in syslog not reading syslog.conf 2015-09-14 10:25:21 #4640 2015-09-14 10:28:38 hydrajump: if you are using stock syslog, it's the busybox implementation. if you want Eric Allman's syslogd, you can find it in inetutils-syslogd 2015-09-14 10:29:17 ScrumpJack: commenting that line works! So alpine users use Allman's syslogd or syslog-ng 2015-09-14 10:30:16 or busybox syslogd 2015-09-14 10:30:52 ok which do you use? 2015-09-14 10:31:06 I don't know one syslog from the other :P 2015-09-14 10:33:28 I use busybox syslogd. I don't it. If you want more config options, inetutils-syslogd is fine. 2015-09-14 10:34:14 syslog-ng adds tcp, and better filtering 2015-09-14 10:34:45 perhaps tls? 2015-09-14 10:35:14 i just need the ability to separate the logs so everything doesn't go into /var/log/messages 2015-09-14 10:35:17 nothing fancy really 2015-09-14 10:35:49 but as was discussed yesterday the busybox syslogd doesn't read syslog.conf unless you run edge 2015-09-14 10:36:04 but I can use rsyslog for now thanks to you ;) 2015-09-14 10:36:06 did we patch it? 2015-09-14 10:36:43 16:50:57 pnutzh4x0r | right, so the current solution is to use another syslog, use busybox from edge, or recompile busybox w/ that config 2015-09-14 10:36:45 | set 2015-09-14 10:37:00 16:45:33 pnutzh4x0r | support for /etc/syslog.conf wasn't added until about a month ago: 2015-09-14 10:37:02 | http://git.alpinelinux.org/cgit/aports/commit/?id=6b80c507b826e38b493422719de9ef55afd4cf84 2015-09-14 10:37:28 ah, cool 2015-09-14 10:37:56 maybe you can help me with my next alpine issue :P 2015-09-14 10:38:03 use inetutil-syslog or syslog-ng. you'll get more help 2015-09-14 10:38:14 this is not syslog related. 2015-09-14 10:38:53 the postfix package created a vmail user, but I need to change the UID of that user. When I try to run `usermod -u 500 vmail` the command usermod is not found 2015-09-14 10:38:59 is this some alpine thing? 2015-09-14 10:39:56 you could just edit the password file, and chmod the files owned by vmail 2015-09-14 10:42:25 there is a usermod utility in the shadow package, but again, that's in edge 2015-09-14 10:50:34 anyone used this? https://www.dyne.org/software/tomb/ 2015-09-14 10:50:46 https://aur.archlinux.org/packages/tomb/ https://aur.archlinux.org/packages/tomb-kdf/ 2015-09-14 10:50:51 nice little wrapper to cryptsetup 2015-09-14 10:51:11 we could easily package it too, as it would certainly not have any issues with musl, (as it's a shell script) 2015-09-14 10:51:21 on top of that all the dependencies are met 2015-09-14 10:51:54 up until now i'd been using an alias in my bashrc 2015-09-14 10:52:02 but i think tomb is a much "safer" and better tested way than that 2015-09-14 10:52:11 least from the sourcecode i've looked at 2015-09-14 10:52:31 looks very elegantly written 2015-09-14 11:28:09 hydrajump: why do you need the vmail user to be uid 500? 2015-09-14 12:00:39 uggedal: dovecot is complaining because it doesn't allow a user with uid below 500. I _could_ change the dovecot setting to allow it, but I assume that that would be a bad security practice? 2015-09-14 12:01:28 uggedal: so the vmail user doesn't have to be uid 500 it could be anything above 500. At the moment the vmail user looks like this as created by the alpine postfix package: 2015-09-14 12:01:36 uid=100(vmail) gid=208(postdrop) groups=208(postdrop),208(postdrop),1002(vmail) 2015-09-14 12:02:41 did someone remove "bc" from the standard installation? 2015-09-14 12:04:16 the alpine iso? 2015-09-14 12:04:34 the default install 2015-09-14 12:04:41 as far as i remember it contained bc 2015-09-14 12:04:51 but last night when i was out in the field and needed it it was not there anymore 2015-09-14 12:05:43 install from which source? the alpine-X.X.X-x86_64.iso? 2015-09-14 12:06:02 yeah 2015-09-14 12:06:04 the normal iso 2015-09-14 12:06:28 or maybe it was a dependency of something that is now gone 2015-09-14 12:06:37 but it was all of the sudden not there anymore 2015-09-14 12:08:58 i don't think dependancies that are removed from a package get uninstalled 2015-09-14 12:12:40 what do you get from apk info -r bc 2015-09-14 12:24:43 nice to read, https://github.com/lfit/itpol/blob/master/linux-workstation-security.md 2015-09-14 12:29:40 just ran an apk update && apk version and these came up: 2015-09-14 12:29:43 musl-1.1.9-r5 < 1.1.11-r2 2015-09-14 12:29:45 musl-utils-1.1.9-r5 < 1.1.11-r2 2015-09-14 12:29:58 are they safe to upgrade? probably a silly question. 2015-09-14 12:32:18 hydrajump: ttp://git.alpinelinux.org/cgit/aports/log/main/musl?h=3.2-stable 2015-09-14 12:55:06 Hello, can you use alpine for the desktop 2015-09-14 12:55:25 toriso: yes 2015-09-14 12:56:16 jvoisin, and it's more secure than other distros? 2015-09-14 12:56:57 define "secure" :p 2015-09-14 12:58:47 jvoisin, that's the big question lol 2015-09-14 13:08:42 toriso: yes, i use it as a desktop at work every day 2015-09-14 13:11:43 but be prepared to tinker :) 2015-09-14 13:14:42 I'm reading it's an embedded system, not really sure what it means 2015-09-14 13:16:08 you can safely ignore that. we've moved on - use alpine linux everywhere you like :) 2015-09-14 13:18:50 the bad news is that it has no thunderbird in its packages ;) 2015-09-14 13:18:52 (yet) 2015-09-14 13:19:25 anyone use Docker? know of any cool Alpine-based images? 2015-09-14 13:19:27 I've been writing a bunch of my own Alpine images and wouldn't mind looking at other people's for ideas 2015-09-14 13:22:43 yah not a huge fan of the docker ecosystem 2015-09-14 13:32:12 anyone using powerline with urxvt? 2015-09-14 13:53:14 cameronmalek: there is a bit of noise about alpine docker images, but LXC seems to be more popular when alpine is a host or guest 2015-09-14 13:54:36 yeah Alpine isn't that popular among Docker-folk. They're mostly in love with debian. 2015-09-14 13:55:57 docker just doesn't do anything special really 2015-09-14 13:57:46 for me, docker doesn't meet the tranparancy and low level playability that I like, where as LXC does 2015-09-14 13:59:10 also, i'm not sure the "images" thing is sustainable long term 2015-09-14 13:59:15 I prefer Docker's community—much more material there. also I'm not aware of any tools like Docker's swarm, compose, or machine or other deployment/remote monitoring tools 2015-09-14 14:00:20 hmm yes seems to be with urxvt 2015-09-14 14:00:22 eg 2015-09-14 14:00:24 echo -e "\xee\x82\xa0\xee\x82\xa1\xee\x82\xa2\xee\x82\xb0\xee\x82\xb1\xee\x82\xb2\xee\x82\xb3" 2015-09-14 14:00:35 seems to output the right powerline font chars in lxterminal 2015-09-14 14:00:39 but it doesn't seem to with urxvt 2015-09-14 14:00:59 and im certainly using a font that supports them 2015-09-14 14:10:20 cameronmalek: I'm not suggesting you change, but I've never needed any feature that I couldn't shell script with LXC. The other day (as publicised here) I had a requirement to deploy 120 alpine lxc guests on an alpine host, start and auto run stuff in all of them. it went fine with just the standard LXC tools, config files and a couple of shell script 2015-09-14 14:11:26 Just looking at the docker ecosystem makes my head spin 2015-09-14 14:12:01 haha yeah it made my head spin too. but I've got it mostly figured out now. 2015-09-14 14:12:28 cameronmalek: I wish I had your patience :) 2015-09-14 14:13:05 you're the one with patience. I learned docker so I didn't have to write shell scripts for every little thing. 2015-09-14 14:13:14 I'm a freelance web developer and I need all the time I can get 2015-09-14 14:13:30 fair enough 2015-09-14 14:14:41 cameronmalek: where does alpine linux come in for you? 2015-09-14 14:15:06 s/come in/make sense/ perhaps 2015-09-14 14:15:08 I don't want to be waiting for 300MB ubuntu images to upload and download 50-100 times a day. that adds up. again, it's about saving time. 2015-09-14 14:15:19 so, size 2015-09-14 14:15:46 and alpine is just simpler when not running into musl libc issues. 2015-09-14 14:16:01 makes sense 2015-09-14 14:58:33 largely confirm what ScrumpJack stated. 2015-09-14 15:00:15 can't you just backup an lxc container to an image just like docker? 2015-09-14 15:00:19 and transfer. 2015-09-14 15:08:19 Sleaker: I cached the apks needed to build a LXC guest, so there was no need to export and transfer. 2015-09-14 15:08:25 also that 2015-09-14 15:08:43 I think someone is providing LXC images somewhere, but i can't remember more 2015-09-14 15:09:24 meh I made an ubuntu one for work, cause we were using vagrant/vbox to build 100% fresh systems 2015-09-14 15:09:44 I re-wrote a base image and gave commands on how to use overlayfs over the top of it to speed thigns up considerably 2015-09-14 15:36:17 hello 2015-09-14 15:36:25 hello 2015-09-14 15:36:31 Sleaker: https://insights.ubuntu.com/2015/06/30/publishing-lxd-images/ 2015-09-14 15:37:03 i've a problem, the last docker package (1.8.2) is segfaulting 2015-09-14 15:37:28 more and more docker questions here. perhaps I should go take a look at docker :) 2015-09-14 15:37:47 it's a good image fo docker :) 2015-09-14 15:38:39 is there a way to get the 1.8.1 package ? 2015-09-14 15:39:25 just installed docker-1.8.1 and it seems to start fine 2015-09-14 15:39:54 docker 1.8.1 works fine 1.8.2 keep segfault 2015-09-14 15:40:05 but i can't find docker 1.8.1 2015-09-14 15:44:42 you could rebuild it 2015-09-14 15:44:59 from source ? 2015-09-14 15:45:39 i will try 2015-09-14 15:46:06 build it into an alpine package i mean 2015-09-14 15:46:31 i never done that but i found that : http://wiki.alpinelinux.org/wiki/Creating_an_Alpine_package 2015-09-14 15:47:18 just docker 1.8.1 for you now. it won't be a signed package so you'll have to grab my key or just --allow-untrusted 2015-09-14 15:47:38 ^building 2015-09-14 15:47:54 ok thanks 2015-09-14 15:48:14 how you do that ? 2015-09-14 15:49:08 habit 2015-09-14 15:49:24 you had the right wiki page though 2015-09-14 15:51:28 drouarb: st.ilet.to/docker-1.8.1-r0.apk 2015-09-14 15:51:51 ty 2015-09-14 15:53:40 uh it keep segfaulting 2015-09-14 15:53:48 :S 2015-09-14 15:54:00 i heate docker 2015-09-14 15:54:04 hate* 2015-09-14 15:54:24 hmm, works here. 2015-09-14 15:55:12 rc-service docker stop 2015-09-14 15:55:24 apk del --purge docker 2015-09-14 15:55:37 it's in a docker container i've built from docker:edge 2015-09-14 15:56:01 you're running docker nested in a docker container? 2015-09-14 15:56:18 yes 2015-09-14 15:56:35 can't help you there, I know nothing about docker 2015-09-14 15:56:41 it's complicated i need to run some container on a paas platform 2015-09-14 15:56:56 what host? 2015-09-14 15:57:02 clever-cloud.com 2015-09-14 15:57:21 i'm mad it worked a week ago 2015-09-14 15:58:06 what host os? 2015-09-14 15:58:57 on my pc it's an archlinux and on server it's exherbo 2015-09-14 15:59:09 but alway in a docker container 2015-09-14 16:00:04 i'm sorry, I'm being a bit dense. what is the host linux distro that ou have access to at clever-cloud? 2015-09-14 16:00:56 you don't have access to the host distro and it's an exherbo 2015-09-14 16:01:10 or is that invisible? you just get to push docker images blindly? 2015-09-14 16:01:25 yes you push and it deploy in the cloud 2015-09-14 16:01:39 you dont have direct access to the host os 2015-09-14 16:01:53 and you want to deploy a docker image with docker inside :) 2015-09-14 16:01:59 yes 2015-09-14 16:03:16 You have : Host -> vm with docker -> my container + my app -> container launched by my app :D 2015-09-14 16:10:23 i have to go i will search tomorrow 2015-09-14 16:10:26 thanks for your help 2015-09-14 16:38:26 hmm im thinking the reason powerline doesn't work in alpine-linux with urxvt is because of a compilation option maybe 2015-09-14 16:40:46 https://i.imgur.com/eNTYT2o.png 2015-09-14 16:41:31 https://dpaste.de/YYSj .Xresources 2015-09-14 16:48:00 https://i.imgur.com/gARepaX.png as you can see here 2015-09-14 16:57:03 what is powerline? 2015-09-14 16:59:31 powerline is basically a python script that can be used to make pretty prompts 2015-09-14 16:59:41 on things like zsh, or vim, tmux or various other tools 2015-09-14 17:00:04 eg https://github.com/powerline/powerline 2015-09-14 17:00:32 https://github.com/bling/vim-airline 2015-09-14 17:00:32 is your TERM ok? I use TERM=rxvt or TERM=rxvt-unicode-256color 2015-09-14 17:00:42 and not xterm-256color 2015-09-14 17:00:48 hmm 2015-09-14 17:01:14 it is currently set to xterm-256color! 2015-09-14 17:01:28 maybe i should try rxvt-unicode-256color 2015-09-14 17:02:13 need to reopen irc 2015-09-14 17:03:03 hmm 2015-09-14 17:03:13 jomat: do you mean in Xresources? 2015-09-14 17:03:34 i changed it to URxvt.termName: rxvt-unicode-256color 2015-09-14 17:03:39 but i don't think that's enough 2015-09-14 17:04:10 Did it affect your TERM variable? (echo $TERM) 2015-09-14 17:04:41 jomat: oh okay i did that 2015-09-14 17:04:42 and no 2015-09-14 17:04:47 user@palanthas  ~  echo $TERM 2015-09-14 17:04:49 rxvt-unicode-256color 2015-09-14 17:04:53 ah ok 2015-09-14 17:04:56 still shows up as the two square symbols 2015-09-14 17:05:01 instead of the full arrows 2015-09-14 17:05:18 btw: do you see those: ▶ 0:irssi ▶ 2015-09-14 17:05:19 i think it might be that our version of urxvt-unicode doesn't have unicode3 2015-09-14 17:05:35 yes i do jomat 2015-09-14 17:05:41 just not the extra powerline symbols 2015-09-14 17:05:51 ok…  are boxes here, too :-/ 2015-09-14 17:05:52 even though i am using a font which supports them 2015-09-14 17:06:11 how does this look for you: 2015-09-14 17:08:42 https://raw.githubusercontent.com/bling/vim-airline/master/doc/airline.txt 2015-09-14 17:08:48 see 2015-09-14 17:08:50 CUSTOMIZATION *airline-customization* 2015-09-14 17:09:22 i. e. this works: let g:airline_right_sep = '◀' 2015-09-14 17:09:35 this is a box: let g:airline_symbols.crypt = '🔒' 2015-09-14 17:10:22 yeah for me it looks like https://i.imgur.com/8q5SVkx.png 2015-09-14 17:10:32 the crypt one does not work that's correct. 2015-09-14 17:10:43 not sure why that is 2015-09-14 17:11:00 the powerline symbols are completely broken here 2015-09-14 17:11:07 nor does the airline_symbols.whitespace 2015-09-14 17:11:12 but the rest are all fine for me 2015-09-14 17:11:16 in xfce4-terminal 2015-09-14 17:11:29 https://github.com/powerline/powerline/issues/453 2015-09-14 17:11:52 this bug seems to indicate urxvt needs to be compiled with --enable-unicode3 2015-09-14 17:13:37 and no, i don't see the LN 2015-09-14 17:13:38 like he does 2015-09-14 17:20:25 yeah im thinking it's that compile option 2015-09-14 17:20:35 ill have to set up an APKBUILD environment 2015-09-14 17:22:01 https://projects.archlinux.org/svntogit/community.git/tree/trunk/PKGBUILD?h=packages/rxvt-unicode 2015-09-14 17:26:10 sillysausage: http://git.alpinelinux.org/cgit/aports/tree/main/rxvt-unicode/APKBUILD 2015-09-14 17:28:51 i've had a look at the different flags as it happens i was doing that 2015-09-14 17:29:17 So yes, there is no enable-unicode3 in our urxvt 2015-09-14 17:29:32 https://dpaste.de/JNj4 2015-09-14 17:29:37 ours differs with those things 2015-09-14 17:29:50 ie thats how it differs from archlinux 2015-09-14 17:29:54 i should find out what those things are 2015-09-14 17:30:04 yes 2015-09-14 17:31:32 hmm http://cvs.schmorp.de/rxvt-unicode/ 2015-09-14 17:50:28 can i build an APK without downloading the whole tree 2015-09-14 17:50:52 if i just want to change one build flag 2015-09-14 17:51:51 eg if i wanted to just get http://git.alpinelinux.org/cgit/aports/plain/main/rxvt-unicode/APKBUILD 2015-09-14 17:52:12 and add --enable-unicode3 \ 2015-09-14 17:53:00 any clue how much space a minimal install of alpine uses on disk installs? 2015-09-14 18:08:34 Diftraku: depends if you need a kernel... then its about a few 100 mb (sth. 200 or 300 afair) 2015-09-14 18:08:53 just a chroot?container-fs is 2015-09-14 18:09:14 just a chroot/container-fs is sth. about 10-20 mb.. 2015-09-14 18:09:25 sillysausage: yes 2015-09-14 18:10:12 sillysausage: mkdir -p main/rxvt-unicode, put it there and run abuild -r 2015-09-14 18:13:12 <_mjones> Diftraku: about 256mb of storage. 2015-09-14 18:13:41 <_mjones> Diftraku: ~100mb are just for kernel modules. The actual kernel is no more than 3-4mb. 2015-09-14 18:14:11 <_mjones> Busybox is similarly tiny. The balance is made up of non-busybox userland. 2015-09-14 18:14:30 <_mjones> These are numbers from x86_64. 2015-09-14 18:16:00 <_mjones> There is an emphasis on avoiding unneeded dependencies because such things can quickly balloon the footprint. 2015-09-14 18:25:50 when i build the package jomat where does it go? 2015-09-14 18:26:07 >>> rxvt-unicode: Package is up to date 2015-09-14 18:26:15 it successfully compiled 2015-09-14 18:26:21 and revision bumped itself i think 2015-09-14 18:26:26 does it automatically install it? 2015-09-14 18:28:19 sillysausage: in ~/packages 2015-09-14 18:29:07 sillysausage: I have an entry in my /etc/apk/repositories: @jomat /home/jomat/packages/testing/ 2015-09-14 18:29:31 <_mjones> That's for the packages you build and sign. 2015-09-14 18:29:35 then my own packages are installable with "apk add something@jomat" 2015-09-14 18:29:39 <_mjones> They're not automatically installed. I don't think. 2015-09-14 18:29:41 oh there it is 2015-09-14 18:29:44 yeah you're right 2015-09-14 18:30:14 sudo apk add rxvt-unicode-9.21-r3.apk 2015-09-14 18:30:15 [sudo] password for user: 2015-09-14 18:30:17 ERROR: rxvt-unicode-9.21-r3.apk: UNTRUSTED signature 2015-09-14 18:30:19 hmm 2015-09-14 18:30:31 can i force the key? 2015-09-14 18:30:33 yes 2015-09-14 18:30:36 i signed it with mine 2015-09-14 18:30:48 --allow-untrusted or you add your key to the keys 2015-09-14 18:30:56 ah okay 2015-09-14 18:31:00 ill just do allow untrusted 2015-09-14 18:31:06 under /etc/apk/keys 2015-09-14 18:31:32 tada! 2015-09-14 18:31:34 that fixed it 2015-09-14 18:31:35 :-D 2015-09-14 18:31:47 Now the font works? 2015-09-14 18:31:48 _mjones: with kernel, for pc engines apu2d and friends :3 2015-09-14 18:32:08 pondering if we should change over to alpine instead of openwrt just for the sane environment 2015-09-14 18:32:54 very tempted to have a diskless install, with a preset profile that gets loaded at boot and use the other medias as storage 2015-09-14 18:34:02 <_mjones> Diftraku: I'm strongly considering that board for an upcoming project. 2015-09-14 18:34:10 it's is very nice 2015-09-14 18:34:27 <_mjones> What part of openwrt are causing you problems? 2015-09-14 18:34:27 especially after I got confirmation we can safely EoL our alix boards 2015-09-14 18:35:12 the build system is a makefile hell, a simple update takes 15 minutes (one change in some custom package) and the overall feel is really kludgy 2015-09-14 18:35:16 <_mjones> It's larger, and doesn't fit in the nice IP65 enclosures offered by mini-box, but is otherwise very attractive. 2015-09-14 18:35:19 like no udev 2015-09-14 18:35:29 but that's not a big issue anymore 2015-09-14 18:35:50 I had to kludge the interface order for alix as it's "reversed" for our enclosure 2015-09-14 18:36:29 we just received a fresh batch of new, smaller rackmount cases :3 2015-09-14 18:36:57 it's a bit smaller than previous ones we had but still better in every aspect 2015-09-14 18:37:26 we basically trimmed the depth a bit while keeping it full width for a rack 2015-09-14 18:37:56 basically, it has stand-offs for the board and modular bits for other stuff 2015-09-14 18:38:07 <_mjones> Diftraku: I take it you're integrating a large number of these? 2015-09-14 18:38:13 very 2015-09-14 18:38:28 this is for an in-production item 2015-09-14 18:38:46 small batches but significant enough to warrant a custom case 2015-09-14 18:38:57 <_mjones> The miniPCI WLAN is quite attractive. There are a couple of interesting ARM boards that might work, including a new crowdfunded one, but nothing with modular 802.11. 2015-09-14 18:39:16 hehe 2015-09-14 18:39:32 we only need the ethernet and internal USB ports 2015-09-14 18:40:05 been pondering on dropping a middle-man on certain installs by using the serial port on the thing instead of serial-to-ethernet converter 2015-09-14 18:40:44 <_mjones> serial-to-ethernet for what? remote console? 2015-09-14 18:41:01 <_mjones> oh, no. Wrong way I think. 2015-09-14 18:41:02 we need to interface with some legacy stuff that talks over serial 2015-09-14 18:41:51 remote console could be one application, we more or less offer a platform to build on 2015-09-14 19:09:34 ScrumpJack/drouarb: working on fixing docker now 2015-09-14 19:10:31 I think it's the go 1.5 which is at fault, not docker 1.8.2. 1.8.1 also segfaults due to MPROTECT 2015-09-14 19:20:06 guess ill put in a bug report to get someone to add --with-unicode3 useflag 2015-09-14 19:20:25 > Diftraku | pondering if we should change over to alpine instead of openwrt just for the sane environment 2015-09-14 19:20:27 i did that 2015-09-14 19:20:43 i wrote a documentation about it http://wiki.alpinelinux.org/wiki/Linux_Router_with_VPN_on_a_Raspberry_Pi 2015-09-14 19:21:09 and i didn't really have any need for a fancy web interface 2015-09-14 19:22:03 though i am thinking of flashing my http://wiki.openwrt.org/toh/tp-link/td-w9980 with openwrt so i can pull SNR/linestats 2015-09-14 19:22:16 but ill still run it in bridge mode behind the raspberry pi 2015-09-14 19:47:55 woah just found this: https://github.com/ServiceManager/ServiceManager 2015-09-14 19:47:57 interesting 2015-09-14 19:52:52 Sleaker: is that serious, and if so, what advantages does it have over runit? over s6? 2015-09-14 19:53:01 no clue 2015-09-14 19:53:06 it popped up on reddit 2015-09-14 19:53:15 the testimonials at the bottom are pretty awesome though 2015-09-14 19:55:13 yeah, hence why I had to ... ask 2015-09-14 19:58:52 haha ammunta 2015-09-14 19:58:55 yaah. 2015-09-14 20:02:24 wow its the latest and greatest, i know a lot of people at work that will be asking to use it 2015-09-14 20:02:33 they love shiny new things 2015-09-14 20:02:49 mongodb is web scale 2015-09-14 20:03:24 it supports sharting and has database impotence 2015-09-14 20:03:36 obviously we use it 2015-09-14 20:07:10 veXx: impotence 2015-09-14 20:07:34 docker segfault on edge should now be fixed 2015-09-14 20:09:25 veXx: you just have to throw more key-words around. 'Web Scale' didn't get hilighted enough... 2015-09-14 20:09:32 if you say it more it will mean something. 2015-09-14 20:13:08 lol 2015-09-14 20:13:17 <_mjones> serious question: anyone know if mongodb3 with wiredtiger is now fsyncing or whatever it needs to do to shed its bad reputation? 2015-09-14 20:14:08 idk but i have to do a full risk assessment of it by the end of the month 2015-09-14 20:14:21 I thought it was impossible for mongod to fix it's actual design. 2015-09-14 20:14:36 lovely... oh and my hands are tied to doing a tabletop assessment... 2015-09-14 20:15:03 <_mjones> VeXx: I've been in that code just enough to know it has a lot of dependencies. Not as many as ceph... 2015-09-14 20:15:40 <_mjones> Sleaker: so I never bothered to look up what was the core criticism that everyone knows. On this subject I am relatively uninformed. 2015-09-14 20:16:02 the funny thing, days before i got the request, i was showing everyone this: https://www.youtube.com/watch?v=b2F-DItXtZs 2015-09-14 20:16:16 <_mjones> But I'm reminded of it whenever I wear my mogodb t-shirt and someone makes a comment. (I haven't ever used mongodb.) 2015-09-14 20:16:28 and wouldnt you know, in the description request... it list mongodb as being web scale 2015-09-14 20:16:38 my jaw hit the floor 2015-09-14 20:17:23 <_mjones> veXx: I think you'll like https://www.youtube.com/watch?v=GYbNNJklCFQ 2015-09-14 20:17:48 ugh 2015-09-14 20:17:52 <_mjones> I need to remember that, though. DNS is web scale; my devs should use it. 2015-09-14 20:18:28 Sleaker: i saw that ServiceManager thing the other day 2015-09-14 20:18:34 <_mjones> Comments are web scale. Better put one or two in before I see this in code review! 2015-09-14 20:18:35 there was a bit of contraversy over the author though 2015-09-14 20:19:11 lol 2015-09-14 20:20:00 you mean this website? http://www.mongodb-is-web-scale.com/ 2015-09-14 20:20:28 thats where i go when i cant listen to the video 2015-09-14 20:20:29 lol 2015-09-14 20:21:18 I lke the response about data being staged. 2015-09-14 20:21:26 if you want fast results and don't care just write to /dev/null! 2015-09-14 20:22:53 so like mongodb's own headline is 'Applications Never Before Possible' 2015-09-14 20:22:56 not even joking 2015-09-14 20:23:59 <_mjones> So web scale. Is that like O(1)? O(n-1)? 2015-09-14 20:24:11 it's O(0) 2015-09-14 20:24:19 you never get the data. 2015-09-14 20:25:13 <_mjones> Ba-dum-dum! 2015-09-14 20:25:43 <_mjones> Big-O isn't about whether I get my data. It's about time or size. ;) 2015-09-14 20:25:52 <_mjones> Will it take zero time to not get my data? 2015-09-14 20:25:56 complexity technicaly 2015-09-14 20:25:58 http://bugs.alpinelinux.org/issues/4641 easy bug to fix ;) 2015-09-14 20:26:06 for someone who can be bothered 2015-09-14 20:26:09 or number of iterations for worst case algorthm. 2015-09-14 20:26:17 I guess that's 'time' 2015-09-14 20:26:31 well more about acess 2015-09-14 20:26:32 depends on how it's being used I guess 2015-09-14 20:26:47 ncopa: seems to be away and now i am too 2015-09-14 20:26:50 sillysausage: some would see --disable-unicode3 as a feature ;) 2015-09-14 21:22:12 https://github.com/alpinelinux/aports/blob/master/main/apk-tools/0001-io-fix-posix_fallocate-failure-handling.patch wait why 2015-09-14 21:26:16 we can't just apply that upstream? 2015-09-14 21:30:59 <_mjones> oh, it's on github now? 2015-09-14 21:31:12 <_mjones> Accepting pull requests? 2015-09-14 21:32:00 <_mjones> That will reduce the entry barriers to make actually submitting my small stack. 2015-09-14 21:32:07 <_mjones> s/make/me/ 2015-09-14 21:33:12 not sure if pull requests are accepted 2015-09-14 21:33:42 seemingly yes 2015-09-14 22:54:42 oooh ammunta cool. 2015-09-14 22:55:27 I still need to get on writing some tools for remastering alpine onto a USB so it functions like an HD install 2015-09-14 23:23:01 speaking of 'web scale' 2015-09-14 23:23:14 this just happened too: https://google.github.io/lovefield 2015-09-14 23:23:25 yup. that's right people. 2015-09-14 23:23:32 google has no written a DB in javascript. 2015-09-14 23:23:34 now* 2015-09-14 23:30:08 <_mjones> I can imagine some uses for that. 2015-09-14 23:30:47 <_mjones> I'm not sure of the significance of the fact that I can imagine some uses for that. 2015-09-15 01:58:58 Looking for results in English? / Change to English / Continuer en Français / Paramètres de langue 2015-09-15 01:59:01 why is it 2015-09-15 01:59:01 french 2015-09-15 06:37:11 hi all, alpine is great, but I do not see any busybox syslog messages in /var/log/messages, anything I might have missed ? Thanks 2015-09-15 06:41:08 what do you mean busybox messages? 2015-09-15 06:41:26 I mean busybox syslog, not syslog-ng 2015-09-15 06:42:10 I did a rc-update add syslog and it shows that syslog is started in /var/log/messages, but no futher loggin happens 2015-09-15 06:42:28 yah busybox doesn't log very much by default 2015-09-15 06:42:45 you have 2 options. update to busybox in edge/main 2015-09-15 06:43:00 and then add a conf to get busybox-syslog to log more 2015-09-15 06:43:07 or install another syslog package 2015-09-15 06:43:57 if update to edge/main syslog, and add a conf, will it log from awall ? or do I need syslog-ng ? 2015-09-15 06:44:05 no clue 2015-09-15 06:44:13 I think it just adds a syslog compatible daemon 2015-09-15 06:44:23 but I don't use busybox syslog so I'm not sure 2015-09-15 06:44:24 no worries, I will give it a go 2015-09-15 06:44:54 thanks for the feedback 2015-09-15 06:44:59 I just know someone had the same issue last night and those were the 2 options :) 2015-09-15 06:46:00 great, I have to say after years of gentoo and arch, alpine is fresh breeze ! 2015-09-15 07:19:43 morning campers 2015-09-15 07:22:45 he, thats actually true 2015-09-15 07:22:49 good morning 2015-09-15 07:24:35 moin 2015-09-15 07:25:01 are you backporting testing patches from grsec? 2015-09-15 08:15:14 Sleaker : got it, awall logs to /var/log/kern.log with syslog-ng, possibly also with busybox syslog. 2015-09-15 08:25:09 it seems busybox syslog does not log awall drops, syslog-ng does 2015-09-15 09:19:35 excuse the stupid question here, but if alpine linux runs in a docker container, the linux kernel isn't patched is it? 2015-09-15 09:19:43 because it runs the host's linux kernel 2015-09-15 09:21:40 are there any limitations to alpine on desktop? poor intel driver support? no browser? 2015-09-15 09:26:29 I use alpine on my work notebook since over a year 2015-09-15 09:27:32 jomat: which browser do you use? 2015-09-15 09:30:35 firefox 2015-09-15 09:33:10 ACTION is still hoping to see thunderbird for alpine... then he will switch the computer at his work to alpine 2015-09-15 09:33:30 Apparently thunderbird is not trivial as it has been omitted. 2015-09-15 09:35:55 slypheed may be an alternative? 2015-09-15 09:36:31 sylpheed 2015-09-15 09:37:39 hendry: alpine as docker container will use kernel from docker host 2015-09-15 09:37:58 limitations for alpine desktop is normally bad closed source support 2015-09-15 09:38:06 so skype, etc normally dont work 2015-09-15 09:38:23 closed source graphics drivers etc does not work 2015-09-15 09:38:24 greguu: People around me have the bad habbit of sending me invitations to meetings in the outlook format. The lightning plugin for thunderbird takes care of it. Also, I would need to get all folders, contacts, etc moved. 2015-09-15 09:38:44 i have used alpine as desktop for many years now 2015-09-15 09:38:49 i use claws-mail 2015-09-15 09:39:19 adobe-flashplayer works kind of 2015-09-15 09:39:45 engblom: i have not had the time to fix thunerbird package 2015-09-15 09:39:46 sorry 2015-09-15 09:40:02 we used to have evolution 2015-09-15 09:40:14 i dont know if it still works 2015-09-15 09:40:32 ncopa: Np. It is your spare time after all, so I can not put any demand 2015-09-15 09:41:22 All I can do is wish things... and if my wishes comes true I do a switch. 2015-09-15 09:41:54 I am running alpine in a VM for testing and yet I have not had any problem with flash-plugin 2015-09-15 09:41:55 you need someone who can help you fulfil that wish 2015-09-15 09:42:19 it is probably not too difficult to fix thunderbird, but it will likely require some time 2015-09-15 09:44:40 ncopa: what about google chrome ? 2015-09-15 09:44:54 we have a pcakge for it 2015-09-15 09:45:02 but i havent used it much 2015-09-15 09:45:25 netflix does not work 2015-09-15 09:45:29 (yet) 2015-09-15 09:45:44 interesting 2015-09-15 09:46:04 what is interesting is that adobe-flashplayer works 2015-09-15 09:46:06 with firefox 2015-09-15 09:46:19 musl libc provides limited gnu libc ABI compat 2015-09-15 09:46:24 did you guys ever thing about doing static builds? doesn't musl facilitate more efficient linking? 2015-09-15 09:46:49 i have thought of static linking 2015-09-15 09:46:58 there was some idea of adding -static subpackages 2015-09-15 09:47:07 so you could build yourself 2015-09-15 09:47:10 generally 2015-09-15 09:47:18 ncopa: I wonder if the flashplugin is statically linked. I also got very suprised when I noticed it worked without any extra tweaking and work 2015-09-15 09:47:52 generally i prefer dynamic linking, because it simplifies maintenance 2015-09-15 09:48:05 if there is a (security) bug fix in a library 2015-09-15 09:48:12 then you only need push the fix once 2015-09-15 09:48:13 I agree with ncopa on this 2015-09-15 09:48:23 if you build statically, you need rebuild it all 2015-09-15 09:48:38 it might be difficult to know what you need to rebuild 2015-09-15 09:48:58 with dynamic linking it is easier to detect what is linked to the given lib 2015-09-15 09:49:43 hendry: http://morpheus.2f30.org/ 2015-09-15 09:49:49 with big libraries you also benefit from shared memory 2015-09-15 09:50:07 hendry: I have not tested that one, but it is from the description a statically linked musl distro 2015-09-15 09:50:20 if the library is tiny or small, then it is probably more efficient with static link 2015-09-15 09:50:34 thanks guys 2015-09-15 09:51:06 alos if you only use one smallish function from lib, then it is more efficient with static link 2015-09-15 09:51:08 another question: how come the docker image is 5mb and in the installed version is >110mb ? 2015-09-15 09:51:21 kernel is huge 2015-09-15 09:51:21 is anyone working on bringing Rust to Alpine? 2015-09-15 09:51:30 vegai: not that i know 2015-09-15 09:51:34 so just the linux kernel? sheesh 2015-09-15 09:51:38 yup 2015-09-15 09:52:04 i wish the kernel would go on a slimming exercise one day. sloc is crazzy 2015-09-15 09:52:36 vegai: While rust is an interesting concept, it is still from my understanding very unstable. The only thing that got stable recently is the api, meaning you probably will be able to compile your code also on future versions of rust. Rust is under heavy development 2015-09-15 09:53:24 engblom: no disagreements there, I suppose 2015-09-15 09:53:40 then addendum question, is anyone working on bringing ghc to Alpine? :) 2015-09-15 09:53:52 i saw someone working on it 2015-09-15 09:53:59 and even provide some kind of patches 2015-09-15 09:54:05 i never had time to look at it 2015-09-15 09:54:21 ghc needs bootstrapping 2015-09-15 09:54:34 yeah 2015-09-15 09:54:34 and is pretty complicated 2015-09-15 09:54:41 indeed 2015-09-15 09:55:10 ghc is one of the things I will miss too when switching to alpine... but it is not critical for me. 2015-09-15 09:55:32 ghc is worse than java 2015-09-15 09:55:32 I am using a long time improved configuration of xmonad 2015-09-15 09:55:50 with java you can bootstrap old java with gcc-java 2015-09-15 09:55:57 you cannot do that with ghc 2015-09-15 09:56:03 it needs to be corsscompiled 2015-09-15 09:56:31 From my understanding you need to walk through really many versions of ghc for bootstrapping from scratch. The openbsd people did it 2015-09-15 09:57:39 my dilemma is that Alpine + ARM seems like a very cute experimental server-side app platform 2015-09-15 09:57:54 but the list of nice programming languages for that combo is a bit limited 2015-09-15 09:58:13 vegai: It recently got openjdk. Clojure is running on openjdk 2015-09-15 09:58:30 vegai: as you ask for rust and ghc, I guess you like functional programming 2015-09-15 09:58:39 we have godo lua support ;) 2015-09-15 09:58:42 good* 2015-09-15 10:00:27 engblom: I like safety and I like efficiency 2015-09-15 10:00:34 vegai: I am quite new to alpine too and not an alpine developer or anything like that. I have tested a bit alpine on arm and the biggest trouble with alpine on arm is that setup-disk is broken for arm. You can only run from memory the image or then hack setup-disk as I did to get a chroot environment 2015-09-15 10:00:56 engblom: I'm just running it on Scaleway, truthfully 2015-09-15 10:01:02 it seems to just work there 2015-09-15 10:03:25 vegai: I quickly checked, erlang seem to exist for arm too. 2015-09-15 10:03:36 vegai: That might be an option for you also 2015-09-15 10:06:58 ACTION really loves Clojure. 2015-09-15 10:10:37 btw: I would love to see alpine for armv5tel as a platform. 2015-09-15 10:13:49 ncopa: any chance you can add this flag and push it http://bugs.alpinelinux.org/issues/4641 2015-09-15 10:14:19 i have verified that it works, (package i compiled manually) 2015-09-15 10:17:08 engblom: how's Clojure static type checking these days? 2015-09-15 10:19:59 vegai: does ocaml work 2015-09-15 10:20:19 I forgot to test 2015-09-15 10:20:44 I don't think there's any reason a priori that it shouldn't but 2015-09-15 10:23:56 ammunta: doesn't seem to be in the repos yet 2015-09-15 10:24:12 oh right, it's only in the testing repo 2015-09-15 10:24:20 yeah, it's in testing 2015-09-15 10:24:23 yeah, I might try that. It wouldn't be a bad choice 2015-09-15 10:24:38 a bit archaic, but bearable :P 2015-09-15 10:24:54 at least it *has* a vaguely useful typesystem 2015-09-15 10:25:07 yeah, and pretty good performance 2015-09-15 10:25:17 although not sure if that's true on ARM 2015-09-15 10:25:29 well you can squeeze good performance out of it 2015-09-15 10:27:13 you could also start from opam if that's your sort of thing 2015-09-15 10:27:36 that should give a p good test of whether it'll work 2015-09-15 12:34:11 <_mjones> mitchty was around yesterday. He packaged ghc successfully for x86_64, but apparently armhf ports needs llvm-3.5 backend while Alpine supports only llvm-3.6. 2015-09-15 12:36:38 <_mjones> I brainstormed a couple of alternatives, but it didn't sound like there were easy answers for ARM. 2015-09-15 12:38:43 <_mjones> greguu: I have an armv5tel box here that I've been contemplating converting to Alpine, but I doubt there will be enthusiasm for such an older port in the tree. 2015-09-15 12:55:17 anyone around with commit access? 2015-09-15 12:58:33 ill make a patch. 2015-09-15 12:59:23 <_mjones> sillysausage: patch for what? 2015-09-15 12:59:38 just add a useflag and rebuild a package I've tested it localyl 2015-09-15 12:59:52 http://bugs.alpinelinux.org/issues/4641 real easy :P 2015-09-15 13:00:11 i just am not a dev so i can't comit 2015-09-15 13:00:12 <_mjones> Oh, right. ;) 2015-09-15 13:00:48 if you've got a copy of the build tree there it would take you less than a minute :P 2015-09-15 13:00:51 im checking one out now 2015-09-15 13:01:02 <_mjones> Patches are submitted on list, and apparently there's a full tree on github that's accepting pull requests. 2015-09-15 13:01:02 but i still wouldn't be able to commit it anyway 2015-09-15 13:01:09 oh 2015-09-15 13:01:17 okay then ill dothat 2015-09-15 13:01:26 <_mjones> I guess I should check site and wiki for updates on that front. 2015-09-15 13:01:53 i was thinking of making an APK for thunderbird 2015-09-15 13:02:02 im not sure why there isn't one there seems there was long time ago 2015-09-15 13:02:20 but i haven't been able to find out if there's some reason its not possible 2015-09-15 13:02:29 <_mjones> Seems it would be welcome. 2015-09-15 13:02:47 maybe it doesn't work with musl? 2015-09-15 13:02:50 <_mjones> But thunderbird is big, and old ports don't always mean doing the current version will be easy. 2015-09-15 13:03:09 <_mjones> You have nothing to lost by trying, but I'm saying don't assume it will be trivial or easy. 2015-09-15 13:03:15 <_mjones> s/lost/lose/ 2015-09-15 13:03:24 yeah im just wondering if there was someone who tried it 2015-09-15 13:04:09 <_mjones> Big c++ packages tend to be difficult, in my limited experience. 2015-09-15 13:04:16 probably. 2015-09-15 13:04:31 it would be nice if there was a good mail client 2015-09-15 13:04:40 claws-mail is okay but it lacks in some areas 2015-09-15 13:04:43 for example caldav/cardav 2015-09-15 13:05:06 and it's RSS implementation gets some dates wrong when used with musl 2015-09-15 13:05:46 <_mjones> hmm. I wonder if that's an RFC2822 issue. 2015-09-15 13:05:55 yeah it is i think 2015-09-15 13:06:01 <_mjones> ACTION mostly only does ISO 8601 dates. 2015-09-15 13:06:04 but basically claws-mail people couldn't be fucked fixing it 2015-09-15 13:06:25 iirc they were using some date option musl doesn't have but glibc does 2015-09-15 13:06:33 <_mjones> Software is terrible. Hardware is worse. Firmware is the worst of all. 2015-09-15 13:06:46 <_mjones> Yeah, sounds plausible. ;) 2015-09-15 13:07:19 <_mjones> One of the downsides of Linux is that there were a lot of naive coders that started on it years ago, and couldn't write portably to save their lives. 2015-09-15 13:07:58 <_mjones> I was there, I ported their stuff. To big-endian, to 64-bit, to not /usr/include/linux directly included. 2015-09-15 13:08:20 im one of those noobs still 2015-09-15 13:08:32 <_mjones> The downside today is that Linux+glibc is such a dominant platform that although code is much better, it still gets very little testing on any other platforms. 2015-09-15 13:08:54 part of the reason i have a alpinelinux vm for development 2015-09-15 13:08:57 so i can use musl 2015-09-15 13:09:18 if it works with musl it will work with glibc :) 2015-09-15 13:09:22 <_mjones> I didn't say I was an expert -- I'm quite a bad coder. I'm just saying I was there and I knew certain things a lot better than did they. 2015-09-15 13:09:50 <_mjones> sillysausage: if it works with musl and glibc it should be awfully clean. 2015-09-15 13:10:10 <_mjones> Put it on one non-Linux platform (not Win32!) and one big-endian and you're golden. ;) 2015-09-15 13:10:27 best i can do is a bi-endian raspberry pi lol 2015-09-15 13:10:32 <_mjones> Even google is afraid of endianness issues in their codebase. 2015-09-15 13:10:50 heh 2015-09-15 13:10:57 <_mjones> I don't think any OSes support booting modern ARM as big-endian, do they? 2015-09-15 13:11:08 i don't really know 2015-09-15 13:11:17 i thought only really old IBM machines were big endian 2015-09-15 13:11:30 iirc some sparcs too 2015-09-15 13:11:37 <_mjones> I was running SPARCs and 64-bit Alphas in the 90s. I had to be able to fix endianness and 64-bit issues or a large percentage of software wouldn't work. 2015-09-15 13:11:48 <_mjones> All SPARCs are big-endian. 2015-09-15 13:11:54 ah yeah. 2015-09-15 13:12:22 <_mjones> Most PPC/Power are big-endian, until the latest Power8/OpenPower which are little and which with Google is experimenting. 2015-09-15 13:12:53 <_mjones> Alphas are little. MIPS are....big. Check Wikipedia on that for caveats. 2015-09-15 13:13:13 <_mjones> Right, there's the ci20 board with MIPS on it, if one was so inclined. 2015-09-15 13:13:44 <_mjones> http://elinux.org/MIPS_Creator_CI20 2015-09-15 13:14:47 <_mjones> I like MIPS and I wish it was as popular as ARM and as fast as Power and as open as RISC-V or something. Unfortuntunately it's not. And that board is no panacea. 2015-09-15 13:14:55 <_mjones> MIPS is owned by the PowerVR people, for one. :( 2015-09-15 13:16:10 <_mjones> Aw hell. Spec sheet says that board is little-endian. 2015-09-15 13:17:32 <_mjones> Damn, I should have invested in MIPS stock a few years ago like I thought. Imagination paid a premium. 2015-09-15 13:18:47 <_mjones> So it appears the classic architectures were MIPS big-endian, but the ISA is bi-endian and people today use it little-endian with Linux. 2015-09-15 13:20:00 mm MIPS will be quite common in routers 2015-09-15 13:20:26 <_mjones> Wireless routers. 2015-09-15 13:20:38 i noticed not many choose arm 2015-09-15 13:20:44 <_mjones> Well actually older Ciscos, also. I used to have a pile of Ciscos with IDT Orions in them. 2015-09-15 13:21:00 mine had a intel strata in it 2015-09-15 13:21:19 <_mjones> Which router? I've never seen an intel strata. 2015-09-15 13:21:28 wait that was the memory 2015-09-15 13:21:32 https://en.wikipedia.org/wiki/StrataFlash 2015-09-15 13:21:32 <_mjones> I don't even know what that is. 2015-09-15 13:21:50 i remember it going on about it when i showed uptime 2015-09-15 13:22:05 (obviously not processor, its just nor ram) 2015-09-15 13:22:07 <_mjones> Interesting. Also, intel was begun as a memory semiconductor company. 2015-09-15 13:23:01 <_mjones> That's weird for it to show an advert when you did uptime. Someone got marketing co-dollars for that. 2015-09-15 13:23:59 <_mjones> Anyway, let me be practical. QEMU will do fine for emulating something big-endian. ;) 2015-09-15 13:24:43 <_mjones> I always thought making code endian-clean was fairly trivial, but I was recently told differently by people with more skill than I. 2015-09-15 13:29:08 brb 2015-09-15 13:42:51 yay thanks ncopa 2015-09-15 14:46:42 so just to explain a bit more re ghc+arm, ghc in general is moving to llvm for things, 8.0 out in january..ish will vendor in llvm, for arm, really llvm is the code gen option, what i need from llvm is simply llc and opt binaries 2015-09-15 14:47:10 but they have to be 3.5 for the current version of ghc 2015-09-15 14:48:16 <_mjones> So actually, that's another obvious option. Ship the armhf port with embedded llvm-3.5 dependencies. 2015-09-15 14:51:42 <_mjones> That's the least-bad option discussed, it seems to me. 2015-09-15 14:53:14 so something like an ghc-llvm subpackage? 2015-09-15 14:55:23 <_mjones> Sounds like it. But only on armhf, I assume. You did say LLVM was better on x86_64, but I don't know if the users prefer that or a smaller package. 2015-09-15 15:00:49 its selectable at runtime 2015-09-15 15:01:06 can do -fllvm to force it 2015-09-15 15:01:15 tends to produce better code in general 2015-09-15 15:03:26 so far 3.7/3.6 work with what will become 8.0, but thats neither here nor there for the moment 2015-09-15 15:04:17 <_mjones> if you had to guess at a timeline when you'd want to package 8, what would be your guess? 2015-09-15 15:09:10 once i get this port finished i'm going to work on getting builds of ghc head, probably weekly for that 2015-09-15 15:09:26 that way it'll be ready straight away 2015-09-15 15:11:08 until i get it compiled on arm though, i say don't worry about it 2015-09-15 15:42:25 _mjones: is anyone using /ghc/ caring about the package size 2015-09-15 15:43:41 _mjones: whaaa 2015-09-15 15:43:49 _mjones: mips is mostly little-endian 2015-09-15 15:44:04 and CI20 is a waste of time for a port because mips32 blows 2015-09-15 15:44:39 that's weird 2015-09-15 15:44:58 did the cert change for https://pkgs.alpinelinux.org/ 2015-09-15 15:45:18 Begins On: 08/21/15 2015-09-15 15:45:25 Expires On: 08/20/16 2015-09-15 15:45:37 SHA-256 fingerprint: E8:75:73:9B:EC:6F:3C:44:0D:54:49:3C:3A:3B:BD:07:6F:EB:56:01:E8:54:36:25:CC:E3:8B:E2:1A:62:AC:D3 2015-09-15 15:45:58 my browser for some reason things that the cert just changed 2015-09-15 15:46:22 Could not verify this certificate because the issuer is unknown 2015-09-15 15:46:28 CN: StartCom Class 1 Primary Intermediate Server CA 2015-09-15 15:47:01 or maybe i hadn't visited it with https in the past 2015-09-15 15:50:00 <_mjones> kaniini: I was thinking about IRIX and back when I spent time worrying about byte-order. I guess I was oblivious to 'mipsel' being the dominant arch for Linux, assuming it is. 2015-09-15 15:51:11 oh, huh it has https 2015-09-15 15:51:46 <_mjones> kaniini: I'd adore a modern unencumbered mips64 implementation. The ISA dates from 1991 so can't fundamentally be encumbered. I guess the Chinese were the closest. 2015-09-15 15:52:25 <_mjones> kaniini: even better if the unencumbered mips64 were competitively fast for a desktop or medium server. 2015-09-15 15:53:34 <_mjones> The Chinese signed a deal with someone -- MIPS, I think, before Imagination bought them. I couldn't help but wonder if that wasn't a substantial mistake. 2015-09-15 15:56:59 i wonder if someone will make their own new arch 2015-09-15 15:57:17 for example in my lifetime i've seen 32bit to amd64 2015-09-15 15:57:34 im wondering if i will see another architecture change in my lifetime 2015-09-15 15:57:40 that isn't arm 2015-09-15 15:58:22 maybe 128bit cpus? would there be any advantage in that? 2015-09-15 15:58:37 <_mjones> There are a number of unencumbered ISAs around now. In fact, some people close to Alpine are reviving SH2 as J2 as we speak. 2015-09-15 15:58:56 <_mjones> sillysausage: bing CHERI; IIRC it's 256bit. 2015-09-15 15:59:26 did you just say bing 2015-09-15 15:59:38 oh wait it is actually called a bing cheri lol 2015-09-15 15:59:51 https://en.wikipedia.org/wiki/Bing_cherry 2015-09-15 15:59:52 <_mjones> http://0pf.org/j-core.html 2015-09-15 15:59:55 wait no that's not what i want 2015-09-15 16:00:13 <_mjones> I say bing ironically. 2015-09-15 16:00:28 so you don't actually work for microsoft? 2015-09-15 16:00:29 <_mjones> Always good for a laugh or an uncomprehending look. 2015-09-15 16:00:37 am i seeing product placement in #alpine-linux? 2015-09-15 16:00:50 next up you'll tell me you think the best computer in the world is a microsoft surface 2015-09-15 16:01:01 and that windows 10 is the best operating system ever made 2015-09-15 16:01:09 <_mjones> algitbot, genericwebsearch cheri 2015-09-15 16:01:16 ^_^ 2015-09-15 16:01:52 <_mjones> Technically, I used to run a search engine. If it still existed, I'd point you there and then you could question my motives. 2015-09-15 16:02:08 what was the search engine? 2015-09-15 16:04:23 <_mjones> It wasn't one you'd ever used, although it was general purpose. I'm going to decline to name it. It literally doesn't exist any longer, if that eliminates some candidates for you. 2015-09-15 16:05:32 <_mjones> bing is shorter than 'google'. And I dislike microsoft, and I have a beard and can be mistaken for a hipster, so I'm going to continue to be ironic, thanks. 2015-09-15 16:26:49 maybe the mill cpu arch, but thats vaporware for the moment 2015-09-15 17:42:05 does anyone know the breakdown of users among x86 & x86_64 for alpine? 2015-09-15 19:09:48 icicled - I think most people use 64 bit 2015-09-15 22:26:38 I got this error just now when trying to compile using clang on armv6: http://txt.shiz.me/MDU2ZDBiOT 2015-09-15 22:26:42 it seems to disappear if i install gcc 2015-09-15 22:26:49 (and still use clang) 2015-09-16 01:56:20 So.. from what I understand alpine doesn't have the kdrive x servers? I am trying to build kdrive on gentoo-musl now but I'm not sure if it is going to work... and all in all I'd prefer tinyx (the 1.2.61 updated one from puppylinux I guess the one from tiny core is GPLv3 meh) 2015-09-16 07:20:00 oh hey, I could just use go (re: earlier language discussion with alpine+arm) :P 2015-09-16 07:20:09 that's a supremely boring language, but works quite well 2015-09-16 07:21:38 the language nerd in me just cannot shut up about its weak points :-/ 2015-09-16 08:06:20 vegai: the first time I heard the intent of go I was like "why does this warrant a new language instead of investing in compiler work on e.g. a ML" 2015-09-16 08:06:42 take that for what you will 2015-09-16 08:11:44 if it's boring and problematic for the inner language nerd and has no point but works well it might explain why it's a common sysadmin language now 2015-09-16 08:12:20 boring and reliable are two of a kind 2015-09-16 08:20:06 tbh go seems like a language suited for being totally confused about systems administration 2015-09-16 08:37:22 hehe 2015-09-16 08:37:50 so far i've also preferred just to do what people need instead of hugging go 2015-09-16 08:38:39 but i use a lot of python and i'm totally not impressed. i'm not a dev and so i'm not going to fall in love with it, havent in the last years and won't in the future 2015-09-16 08:39:25 too far away from what i need to do and too big lack of error handling 2015-09-16 08:39:32 rather going with shell state machines 2015-09-16 09:45:52 doing some maintenace for pkgs.a.o 2015-09-16 10:58:42 pkgs.a.o has been updated. you can now select on maintainer. 2015-09-16 11:08:00 neat 2015-09-16 12:21:11 clandmeter: would be cool to get apk to pull info from pkgs.a.o 2015-09-16 12:22:10 at the moment i have to use wget 2015-09-16 12:22:16 apkfind() {wget -q --no-check-certificate -O - https://pkgs.alpinelinux.org/contents\?filename=$1 | sed -e 's/<[^>]*>//g' -e '1,89d' -e '/Copy/d' | xargs -n 4 } 2015-09-16 12:30:23 clandmeter: how does pkgs.a.o get the package info (like file lists)? 2015-09-16 12:31:11 uggedal: good question. we could put *that* into apk :) 2015-09-16 12:44:04 i think i might create my first APK 2015-09-16 12:44:31 https://www.dyne.org/software/tomb/ looks easy enough and we might like it 2015-09-16 12:46:12 hmm although 2015-09-16 12:46:17 when trying it manually i saw this: 2015-09-16 12:50:36 https://dpaste.de/GvbZ 2015-09-16 12:50:45 user@palanthas:~ $ tomb open secret.tomb -k secret.tomb.key 2015-09-16 12:50:47 tomb . Commanded to open tomb secret.tomb 2015-09-16 12:50:49 mount: unrecognized option: l 2015-09-16 12:50:53 so i'm guessing, busybox's mount doesn't have an l option 2015-09-16 12:52:28 sillysausage: use the function apkfind above to look for the package that contains a mount that supports l 2015-09-16 12:55:03 its a neat little script i kind of had aliases in my bashrc 2015-09-16 12:55:07 but i dont really like that 2015-09-16 12:55:23 https://github.com/dyne/Tomb/blob/master/tomb is the script 2015-09-16 12:55:31 uggedal: i generate a json and its avialable on nl mirror 2015-09-16 12:55:59 i also don't really like password managers 2015-09-16 12:56:13 because i find those often don't allow you to keep other information like backup codes if using two-factor auth 2015-09-16 12:56:25 also they are obvious what it is you're trying to hide 2015-09-16 12:56:45 where as a non-descript encrypted file draws much less attention than "passwords.db" 2015-09-16 12:57:19 for password files and other sensitive files i use encfs 2015-09-16 12:58:18 encfs has a few drawbacks 2015-09-16 12:58:27 sEncFS which implements weaker encryption than dm-crypt and it doesn’t promotes the separated storage of keys. 2015-09-16 12:59:57 the other reason i do not like encfs is because: 2015-09-16 13:00:00 This does create a few disadvantages, though; because the encrypted files are not stored in their own file, someone who obtains access to the system can still see the underlying directory structure, the number of files, their sizes and when they were modified. They cannot see the contents, however. 2015-09-16 13:00:01 This particular method of securing data is obviously not perfect, but there are situations in which it is useful. 2015-09-16 13:01:08 i'm not bothered about the files metadata 2015-09-16 13:01:19 ah where as i am :P 2015-09-16 13:01:53 the GNU mount command isn't a part of coreutils? 2015-09-16 13:02:05 apkfind() {wget -q --no-check-certificate -O - https://pkgs.alpinelinux.org/contents\?filename=$1 | sed -e 's/<[^>]*>//g' -e '1,89d' -e '/Copy/d' | xargs -n 4 } 2015-09-16 13:02:13 The option -l adds the labels in this listing. 2015-09-16 13:02:18 use that laziness 2015-09-16 13:02:34 oh i didn't know about that wone :) 2015-09-16 13:03:35 how does that even work 2015-09-16 13:03:59 for zsh 2015-09-16 13:04:05 adapt for your shell :) 2015-09-16 13:04:09 as it happens i am using zsh 2015-09-16 13:04:18 lol. 2015-09-16 13:04:19 put it in your .zshrc 2015-09-16 13:04:22 oh. 2015-09-16 13:04:58 reload your zshrc (source .zshrc) 2015-09-16 13:05:07 yeah i did that. 2015-09-16 13:05:08 apkfind mount 2015-09-16 13:05:27 it happens it's in util-linux i discovered that 2015-09-16 13:05:33 but it seems that doesn't quite work 2015-09-16 13:05:38 the apkfind() thing 2015-09-16 13:05:46 ok cool. what happened? 2015-09-16 13:06:00 do i do apkfind() mount? 2015-09-16 13:06:05 or just "apkfind mount" 2015-09-16 13:06:18 no, just apkfind mount 2015-09-16 13:06:43 well that does this: https://dpaste.de/q5xL 2015-09-16 13:07:02 apkfind() {wget -q --no-check-certificate -O - https://pkgs.alpinelinux.org/contents\?filename\=$1 | sed -e 's/<[^>]*>//g' -e '1,89d' -e '/Copy/d' | xargs -n 4 } 2015-09-16 13:07:12 i put that in the end of my .zshrc and did source .zshrc to reload it 2015-09-16 13:07:34 heh, wierd. it ran mount 2015-09-16 13:07:44 maybe i found a bug :) 2015-09-16 13:08:29 also tada! 2015-09-16 13:08:33 it worked, (the tomb script) 2015-09-16 13:08:33 can you do apk add sed wget 2015-09-16 13:08:44 it just needed util-linux for mount instead of busybox's mount 2015-09-16 13:09:02 yes sed is installed so is gnu wget 2015-09-16 13:09:10 wait i have busybox sed 2015-09-16 13:09:26 make sure that apkfind() function is all on one line 2015-09-16 13:09:31 in your zshrc 2015-09-16 13:09:32 it is 2015-09-16 13:09:48 94 apkfind() {wget -q --no-check-certificate -O - https://pkgs.alpinelinux.org/contents\?filename\=$1 | sed -e 's/<[^>]*>//g' -e '1,89d' -e '/Copy/d' | xargs -n 4 } 2015-09-16 13:09:51 (line 94) 2015-09-16 13:10:33 try wget -q --no-check-certificate -O - https://pkgs.alpinelinux.org/contents\?filename\=mount | sed -e 's/<[^>]*>//g' -e '1,89d' -e '/Copy/d' | xargs -n 4 2015-09-16 13:11:59 $ apkfind 2015-09-16 13:12:01 File Package Repository Architecture 2015-09-16 13:12:03 /bin//mount util-linux main x86_64 2015-09-16 13:12:05 /usr/share/bash-completion/completions//mount bash-completion main x86_64 2015-09-16 13:12:07 yes that worked 2015-09-16 13:12:27 ok, so your shell isn't executing the function 2015-09-16 13:12:57 can you run echo $SHELL ? 2015-09-16 13:13:03 echo $SHELL 2015-09-16 13:13:07 $ echo $SHELL 2015-09-16 13:13:09 /bin/zsh 2015-09-16 13:13:21 here? -> user@palanthas:~ $ 2015-09-16 13:13:22 it's zsh 5.0.7 (x86_64-alpine-linux-musl) 2015-09-16 13:13:32 yeah on that machine 2015-09-16 13:13:50 open an new terminal any try 2015-09-16 13:14:00 /bin/zsh 2015-09-16 13:14:02 user@palanthas:~ $ 2015-09-16 13:14:22 in /etc/passwd :P user:x:1000:1000:Linux User,,,:/home/user:/bin/zsh 2015-09-16 13:15:58 copy paste this: echo "alias moo=ls" >> .zshrc && source .zshrc && moo 2015-09-16 13:17:27 does it list files in your current directory? 2015-09-16 13:17:32 $ echo "alias moo=ls" >> .zshrc && source .zshrc && moo 2015-09-16 13:17:34 /bin/zsh 2015-09-16 13:17:36 zsh: command not found: moo 2015-09-16 13:17:40 hmm 2015-09-16 13:17:59 or did you want me to put that in .zshrc 2015-09-16 13:18:00 http://nl.alpinelinux.org/alpine/filelist/$repo-$arch.json.gz 2015-09-16 13:18:16 for anybody who is interested in apk filelist 2015-09-16 13:18:21 sillysausage: it's already there if you copy/pasted that line 2015-09-16 13:18:32 that's true 2015-09-16 13:19:01 your .zshrc file isn't getting sourced (reloaded) 2015-09-16 13:19:35 yeah that maybe because a urxvt is already open 2015-09-16 13:19:40 ie with irc in it 2015-09-16 13:19:43 and they are all daemonized 2015-09-16 13:19:55 manually sourcing it fixes it though 2015-09-16 13:20:15 ok, now try apkfind mount 2015-09-16 13:20:50 and yes it works now 2015-09-16 13:21:21 phew! 2015-09-16 13:21:22 thanks! 2015-09-16 13:21:33 zap the moo and moo2 aliases 2015-09-16 13:21:35 im not sure what was going on there 2015-09-16 13:21:37 yeah. 2015-09-16 13:22:01 let us know if you have issues packaging tomb 2015-09-16 13:22:17 ah shouldn't it just needs util-linux and psmisc which we have! 2015-09-16 13:22:42 because it seems the slam option (which unmounts and kills things using things inside your tomb) uses fuser -M 2015-09-16 13:22:49 that's a feature of the fuser included with psmisc 2015-09-16 13:23:06 (by the way, i'm just about to setup an encrypted disk to mount on /home - what does tomb help me with?) 2015-09-16 13:23:28 tomb is for creating container dmcrypt containers specifically 2015-09-16 13:23:41 if that encrypted disk is a physical disk you would not want tomb 2015-09-16 13:23:45 what is a container dmcrypt container? 2015-09-16 13:23:46 you would want to do that using crypttab 2015-09-16 13:23:48 yes 2015-09-16 13:23:52 tombs are dmcrypt containers 2015-09-16 13:23:56 not physical devices 2015-09-16 13:24:11 basically think of them like truecrypt containers 2015-09-16 13:24:11 what sort of containers? 2015-09-16 13:24:16 except its a simple zsh cript 2015-09-16 13:24:17 *script 2015-09-16 13:24:31 dmcrypt with ext4 (by default) 2015-09-16 13:24:48 secret.tomb: LUKS encrypted file, ver 1 [aes, xts-plain64:sha256, sha1] UUID: f4bd7826-40f9-4219-b8d0-32b74a98a22c 2015-09-16 13:24:50 eg 2015-09-16 13:24:55 container as in virtual machine/chroot/lxc/docker etc or container as in bucker 2015-09-16 13:24:58 bucket 2015-09-16 13:25:22 well if you want that mounted at boot you're probably going to be using cryptab 2015-09-16 13:25:26 as for having multiple users 2015-09-16 13:25:30 using multiple home directories 2015-09-16 13:25:32 with different passwords 2015-09-16 13:25:39 ecryptfs is probably a better solution in that case 2015-09-16 13:26:16 tombs are designed to be small portable encrypted containers that's all really. 2015-09-16 13:32:37 clandmeter: thanks 2015-09-16 13:51:30 hmm while im at it i should build steghide 2015-09-16 13:52:12 :( which doesn't build against muslk 2015-09-16 13:52:14 :( which doesn't build against musl 2015-09-16 13:52:41 https://dpaste.de/T73a 2015-09-16 13:52:55 AuFile.cc:46:9: warning: deleting object of abstract class type 'AudioData' which has non-virtual destructor will cause undefined behaviour [-Wdelete-non-virtual-dtor] 2015-09-16 13:52:57 delete Data ; 2015-09-16 13:52:59 hmm 2015-09-16 13:53:29 im probably going to need a patch to get that to compile 2015-09-16 13:54:08 ScrumpyJack: i decided i'd make a steghide apk first because tomb can optionally depend on that 2015-09-16 13:54:25 http://steghide.sourceforge.net/download.php 2015-09-16 13:55:34 i'm not actually sure why that doesn't work 2015-09-16 13:56:39 anyway this is C++ 2015-09-16 13:57:31 ok what's not working? 2015-09-16 13:58:06 oh this source im trying to compile 2015-09-16 13:58:15 basically tomb lets you create a key file 2015-09-16 13:58:26 and you can optionally hide it in a jpeg image with steghide 2015-09-16 13:58:56 (all the deps for steghide can be met) libjpeg-turbo-dev libmcrypt-dev zlib-dev mhash-dev 2015-09-16 13:58:58 so i figured why not 2015-09-16 13:59:01 are you using the abuild tool? 2015-09-16 13:59:13 no, i'm compiling it manually first in my home directory 2015-09-16 13:59:16 to see that it works 2015-09-16 13:59:35 i suspect the compiler problem has a patch im hunting around other distributions for it 2015-09-16 13:59:40 it could be because the source is quite old 2015-09-16 13:59:44 http://steghide.sourceforge.net/download.php 2015-09-16 14:00:15 when i tried to make it locally i got this https://dpaste.de/T73a 2015-09-16 14:00:18 i'd just use abuild directly 2015-09-16 14:01:18 ah yes https://projects.archlinux.org/svntogit/community.git/tree/trunk?h=packages/steghide 2015-09-16 14:01:38 https://projects.archlinux.org/svntogit/community.git/tree/trunk/gcc-4.2.patch?h=packages/steghide looks like that will fix it 2015-09-16 14:02:08 also check out viod linux and gentoo for patches 2015-09-16 14:02:18 true true 2015-09-16 14:03:48 app-crypt/steghide looks like it got removed 2015-09-16 14:03:53 from gentoo at some point 2015-09-16 14:04:14 https://sources.gentoo.org/cgi-bin/viewvc.cgi/gentoo-x86/app-crypt/steghide/files/?hideattic=0 2015-09-16 14:04:18 but there are patches t here 2015-09-16 14:05:44 looks like someone couldn't be bothered patching it 2015-09-16 14:05:56 but it works with archlinux so that must mean it works with recent g++ 2015-09-16 14:06:16 https://bugs.gentoo.org/show_bug.cgi?id=319679 was removed some few years ago 2015-09-16 14:07:00 void linux doesn't have the package 2015-09-16 14:07:33 debian does though 2015-09-16 14:07:42 https://packages.debian.org/jessie/steghide 2015-09-16 14:10:38 looks like debian puts all it's patches in one great big file 2015-09-16 14:10:57 their patch looks pretty huge 2015-09-16 14:17:25 has someone of you managed to be connected via wifi and then create a hotspot with that connection over the same wifi card? 2015-09-16 14:17:34 is something like that even possible? 2015-09-16 14:27:48 leo-unglaub: no, if should be possible to configure until you put the wifi device into master mode when loading hostapd 2015-09-16 14:28:07 then if it worked you can connect to it :) 2015-09-16 14:37:19 it seems the individual patches there https://projects.archlinux.org/svntogit/community.git/tree/trunk?h=packages/steghide 2015-09-16 14:37:24 are actually the ones in the debian diff 2015-09-16 14:37:32 but the debian diff does a pile of other things too 2015-09-16 14:39:27 http://manpages.ubuntu.com/manpages/trusty/man1/diff2patches.1.html 2015-09-16 14:39:35 hmm i don't have a debian machine right around this moment :( 2015-09-16 15:27:04 okay ScrumpyJack so i found the two patches 2015-09-16 15:27:09 https://sources.debian.net/src/steghide/0.5.1-9.1/debian/patches/build-with-gcc-4.patch/ 2015-09-16 15:27:11 debian uses that one 2015-09-16 15:27:15 which patches a few other files 2015-09-16 15:27:24 which are not patched in archlinux https://projects.archlinux.org/svntogit/community.git/tree/trunk/gcc-4.2.patch?h=packages/steghide 2015-09-16 15:27:37 but i don't really know enough about C++ to know what is going on 2015-09-16 15:27:46 im leaning towards taking debian's patch because i trust they got it right 2015-09-16 15:28:38 the archlinux one was quite some time ago 2007-06-28 and probably taken from gentoo when it had the package 2015-09-16 16:05:55 hello 2015-09-16 16:07:05 i am in need of help (preferably not from a bot :). i am trying to simply run a static binary that i have dropped into /usr/local/bin and everytime i run, i get "bash: /usr/local/bin/gosu: No such file or directory" 2015-09-16 16:07:40 my PATH is correct, bash has not acted up otherwise, and i have tried calling the binary directly 2015-09-16 16:08:15 being static, there are no runtime dependencies, so i am kinda going nuts here 2015-09-16 16:09:09 bash is not installed by default. ash is used with AL. 2015-09-16 16:09:37 Jean-Scotch: I understand that. I have taken the step of installing it and altering my default shell 2015-09-16 16:10:08 my apologies if that is pertinent, just trying alpine for the first time today 2015-09-16 16:10:51 cornfeedhobo: ok the first thing there: rather do not change the default shell. but this isn't likely what you get that error for 2015-09-16 16:10:56 i can't make sense of gosu 2015-09-16 16:11:15 (you can fire off a bash via .profile if it's a interactive session) 2015-09-16 16:11:16 Jean-Scotch: just to be sure, i actually returned my shell back to ash, and it gave the same error. let me try once more. brb 2015-09-16 16:14:34 doesn't that error mean that /usr/local/bin/gosu cannot be found? 2015-09-16 16:14:42 gosu is a jvm language IIRC 2015-09-16 16:15:03 and it's not in aports 2015-09-16 16:24:48 icicled: it is a su replacement written in GO. mostly (if not only) used for docker https://github.com/tianon/gosu 2015-09-16 16:26:16 hooboy, I love how everything is re-invented again 2015-09-16 16:26:25 cornfeedhobo, is that app in /usr/local/bin ? 2015-09-16 16:27:23 icicled: hah, yea, that was my first thought too. i am just testing the waters so I am trying to keep my opinions down for the time being :p 2015-09-16 16:27:44 you can always use strace to figure out what is trying to be called 2015-09-16 16:29:09 https://ncry.pt/p/3UAn#OIv7HI4j4_y_7j2Rx2XIpejwQYmDn_APtng-ESgsUdU 2015-09-16 16:29:32 file /usr/local/bin/gosu 2015-09-16 16:29:44 it might be a different abi 2015-09-16 16:29:50 x86 v x86_64 2015-09-16 16:29:56 hmmm, indeed file does say not found 2015-09-16 16:29:58 hmmmm 2015-09-16 16:30:24 what's with the 1000 owner? 2015-09-16 16:33:16 unionfs to a host. when i attempt to build locally, i hit a wall there as well. you may be right that i compiled for the wrong architecture :( 2015-09-16 16:33:57 correct me if i am wrong, but unistd.h is apart of glibc, right? 2015-09-16 16:34:51 it should be provided 2015-09-16 16:34:55 oh, hm. i see it is in musl as well. 2015-09-16 16:36:08 well then. big facepalm there. 2015-09-16 16:36:34 indeed i can just compile natively when i remember to install musl-dev 2015-09-16 16:36:42 le sigh. sorry for the disruption 2015-09-16 16:37:39 all good 2015-09-16 16:37:52 IIRC go can compile cross platform with a flag 2015-09-16 16:38:06 you can just do it on the host 2015-09-16 16:38:20 it produces static binaries IIRC 2015-09-16 16:38:24 ya, that is what i am trying to look at now, because the two appear identical 2015-09-16 16:38:28 correct 2015-09-16 16:51:01 now i am just interested for knowledge's sake. the architectures are the same, they even run the same kernel version (same commit even). 2015-09-16 16:51:25 the only difference i can surmise is musl, but that shouldn't be a concern with static compiles 2015-09-16 17:36:20 cornfeedhobo: did you get gosu working? 2015-09-16 17:38:40 ScrumpyJack: yes, thanks. but i am still confused as to why compiling on my Suse 13.2 x86_64 host didn't make a binary that was able to work within Alpine 3.2. I think that is something i will need to address another day, and probably in #go-nuts 2015-09-16 17:43:11 icicled: did you submit a *new* APKBUILD file for libasr, or a patch of the existing? 2015-09-16 17:49:32 patch 2015-09-16 17:49:40 that fixes it 2015-09-16 17:49:43 and makes it work :) 2015-09-16 17:58:41 \o/ 2015-09-16 17:59:13 let me know when it gets approved, i'm eager to test opensmtpd 2015-09-16 18:02:01 they're both in testing 2015-09-16 18:02:03 have at it! 2015-09-16 18:02:22 I've been running it for the last week or so with great success 2015-09-16 18:02:26 along with dkimproxy 2015-09-16 18:06:46 i'm play with it at work tomorrow morning :) 2015-09-16 18:08:16 i'll play, even 2015-09-16 18:55:31 hmm, i'll need to learn about apk 2015-09-16 19:16:50 cornfeedhobo: did you compile with muslc? 2015-09-16 19:17:00 or with something else? 2015-09-16 19:17:49 oh you already talked about that 2015-09-16 19:18:32 muslc looks almost like music in certain fonts i just noticed 2015-09-16 19:21:22 Sleaker: i first compiled on a suse 13.2 host with glibc, and mounted it inside a docker container. then i compiled within said docker container instead, which finally worked. still completely lost, but i have moved on for now 2015-09-16 19:21:39 kk 2015-09-16 19:21:45 well atleast the second one worked 2015-09-16 20:23:24 anyone know if the Alpine mirrors respond to a common virtual host, such as mirrors.alpinelinux.org? 2015-09-16 20:26:19 i can't find general information on the mirrors, as the mailing list seems to link to incorrect posts in search results :\ 2015-09-16 21:17:53 about to try out alpine-linux in virtualbox 2015-09-16 21:18:16 though I am already seriously thinking about switching :D 2015-09-16 21:18:58 kind of torn a little bit between nixos and alpine 2015-09-16 21:24:33 as a host OS? 2015-09-16 21:26:10 andyshinn: is there any particular reason you want such a thing? 2015-09-16 21:26:16 just curious 2015-09-16 21:27:04 cornfeedhobo: go binaries built with cgo are actually linked to the system libc 2015-09-16 21:27:09 and not static by default 2015-09-16 21:27:30 the ability to have a CNAME in front of a couple mirrors for round robin DNS, and to also do some failover with CDN (most CDN i play with pass the frontend domain name to the origins, so they need to respond on something other than their host name as well) 2015-09-16 21:28:35 Shiz: no shit! that was the problem 2015-09-16 21:28:46 :themoreyouknow: 2015-09-16 21:28:54 Shiz: thanks! 2015-09-16 21:29:13 i wish i had known that before 2015-09-16 21:29:16 a simple start is something like mirrors..alpinelinux.org, and opens possibilities for GeoDNS where there might be a couple DNS servers (one in each region) handling the CNAMEs for mirrors.a.o, handing out the closest one per that region 2015-09-16 21:29:21 there's a flag you can pass to the compilation process to make it so 2015-09-16 21:29:24 ya 2015-09-16 21:29:27 i t hink it was '-extldflags -static' 2015-09-16 21:29:36 thanks. very good to know 2015-09-16 21:29:57 i ran into the same while compiling docker machine for an alpine container 2015-09-16 21:30:01 but the latter requires at least some increased maintenance and complexity overhead 2015-09-16 21:31:33 cornfeedhobo: http://blog.hashbangbash.com/2014/04/linking-golang-statically/ 2015-09-16 21:31:51 this is where i got the intel from, it worked for docker machine at least 2015-09-16 21:32:31 but it might not always work because glibc + static linking = problems sometimes 2015-09-16 21:32:37 especially if your application involves networking 2015-09-16 21:32:46 (musl on the other hand links beautifully statically) 2015-09-16 21:34:00 i am actually thinking that for anything i am "distributing", i would probably rather compile inside alpine containers. i am going to doctor up something to automate this. 2015-09-16 21:34:19 because exactly what you have pointed out 2015-09-16 21:34:28 that's what i do these days 2015-09-16 21:34:33 nice 2015-09-16 21:34:39 well not in containers since i run alpine as main on some servers 2015-09-16 21:34:45 but yes, it makes universal binaries for linux a lot easier 2015-09-16 21:38:04 cornfeedhobo: if you want to build in Alpine container, try out gliderlabs/alpine, i just pushed a package CDN addition which greatly speeds up package downloads 2015-09-16 21:38:15 :) 2015-09-16 21:39:19 andyshinn: i saw that one. the build looked complex so i went with the "official" one at docker's hub, but i really only gave it a glance. i'll check it out again 2015-09-16 21:39:59 hehe, well... just FYI, we maintain the official one, with the same builder :p 2015-09-16 21:40:27 https://github.com/docker-library/official-images/blob/master/library/alpine#L9 2015-09-16 21:41:13 but you are correct, the official one is kept pristine and follows upstream 2015-09-16 21:53:43 intended way to create a user on a new alpine-linux system 2015-09-16 21:53:45 ? 2015-09-16 21:57:57 myztic: i'm pretty sure it is just "adduser " 2015-09-16 21:59:02 andyshinn, ty 2015-09-16 23:58:57 pretty awesome job on documentation <3 :) 2015-09-17 01:01:09 i am working on an anycast mirror for alpine updates, actually. 2015-09-17 03:29:11 i'm working on a Fastly CDN setup off the main nl host with failover to lax-noc, is Alpine interested in making this an official mirror? 2015-09-17 03:29:38 it is specifically for the Alpine Linux Docker image, but there is no reason it couldn't be used everywhere 2015-09-17 03:31:51 i didn't realize Alpine controlled IP space to do prefix announces in multiple locations 2015-09-17 03:32:47 ? 2015-09-17 03:38:44 i was under the impression you needed to announce IP prefixes with eBGP in multiple geos to do anycast type hosting 2015-09-17 03:40:59 oh 2015-09-17 03:41:12 i dunno 2015-09-17 04:26:43 Gidday all :) 2015-09-17 04:43:46 andyshinn: incidentally, is the builder image published 2015-09-17 04:44:47 no, not separately, are you looking to use it on its own? 2015-09-17 04:45:56 well, with -p implemented it's actually something that might be useful alone 2015-09-17 04:50:29 I was planning to use it, anyway 2015-09-17 04:53:25 yea, i'll look into separating it out into its own repo tomorrow, can you file an issue for it so i don't forget? shouldn't be too difficult to do 2015-09-17 05:15:02 andyshinn: done 2015-09-17 05:15:12 https://github.com/muhmuhten/docker-alpine/tree/alpine-builder p straightforward tbh 2015-09-17 05:35:24 is there a place or way to browse APKBUILD files that go into the current tree? e.g. nginx 2015-09-17 05:35:42 into the current tree? 2015-09-17 05:36:03 <_ikke_> cornfeedhobo: http://git.alpinelinux.org/cgit/aports/ ? 2015-09-17 05:37:45 ammunta: i could be using the wrong term. i saw the word "aports" and assumed terminology. "package repository"? 2015-09-17 05:38:01 _ikke_: thanks :) 2015-09-17 05:38:12 there's the aports repo, yeah 2015-09-17 05:38:27 if you want to look at stable, there are tags 2015-09-17 05:39:42 ammunta: tags? 2015-09-17 05:40:09 git tags? 2015-09-17 05:40:21 there are git tags and there are the stable branches 2015-09-17 05:40:30 oh dear algitbot is a bit 2015-09-17 05:41:16 in case you're interested in those 2015-09-17 05:41:30 edge basically tracks master 2015-09-17 05:44:31 i am a bit confused. what repo are you referring to? git://dev.alpinelinux.org/aports ? 2015-09-17 06:28:58 Hi. Are package downloads for alpine edge really slow for anyone else right now, or is it just me? 2015-09-17 06:29:14 it's been struggling ~5min to download perl, at 20kB/s 2015-09-17 06:32:20 danderson, which mirror are you using? 2015-09-17 06:32:28 http://dl-4.alpinelinux.org/alpine/edge/main/ 2015-09-17 06:33:46 ... but that's a good point, I should just switch mirrors and see if it still sucks 2015-09-17 06:33:58 try another one 2015-09-17 06:33:59 yes 2015-09-17 06:34:01 exactly 2015-09-17 06:34:28 should have thought of that. Derp. 2015-09-17 06:34:29 Thanks :) 2015-09-17 06:34:40 yw 2015-09-17 06:36:08 yeah, there we go, that's better. 2015-09-17 07:59:20 Hi 2015-09-17 08:01:38 I'm trying to launch the iso with KVM trough a ssh console. Previously I had to issue "grsec nomodeset" at boot prompt to keep the display. But it doesn't work with the lastest stable iso. Am I doing something wrong? 2015-09-17 08:15:51 can you use qemu? i pretty sure that works 2015-09-17 08:16:26 also, test the non grsec kernel, in alpine-vanilla 2015-09-17 08:17:59 hmm 2015-09-17 08:19:09 is nomodeset actually set? 2015-09-17 08:27:37 I use AL with qemu from last december... 2015-09-17 08:28:05 when I say KVM, I mean "qemu -enable-kvm" 2015-09-17 08:28:57 and I did only use the standard (with grsec) iso till now 2015-09-17 08:29:15 what's your graphics card? nvidia? 2015-09-17 08:29:45 I have about 100 AL boxes (baremetal and KVM) running. This time I try it on my laptop with an intel GC 2015-09-17 08:30:59 my laptop is running AL without graphical interface (lastest 3.2.3 up-to-date) 2015-09-17 08:31:28 and it's on that laptop that you're trying to qemu boot the iso? 2015-09-17 08:32:18 the laptop is in another room, that's why I access it with ssh. Later on, I will reproduce the same setup on some remote server... 2015-09-17 08:32:21 yes 2015-09-17 08:33:13 I would like to simulate a bunch of AL boxes to prepare some orchestration tools for my rack... 2015-09-17 08:35:02 whats your qemu command? 2015-09-17 08:43:14 qemu-system-x86_64 -name darkmoon -m 1024 -pidfile /var/run/kvm/toto.pid -cdrom /srv/alpine-3.2.3-x86_64.iso -drive file=/srv/toto,if=virtio -enable-kvm -curses -boot d 2015-09-17 08:52:05 try adding -vga std 2015-09-17 08:52:51 same :( 2015-09-17 08:53:06 -vga cirrus? 2015-09-17 08:53:48 that does the trick! 2015-09-17 08:54:02 also, what's this? -drive file=/srv/toto,if=virtio 2015-09-17 08:54:02 strange as I never had to do that previously! 2015-09-17 08:54:22 /srv/toto is a qcow image to install AL to it 2015-09-17 08:55:00 you shouldn't need -drive file= and if=virtio these days 2015-09-17 08:55:19 just append the path to your qcow2 2015-09-17 08:56:27 Ha! need to upgrade my knowledge it seems ;) I use to use NBD image most often... 2015-09-17 08:58:33 use virtio for paravirtualisation if things are slow, which could be the case on your laptop 2015-09-17 08:58:44 and hope to be able to use RBD soon 2015-09-17 09:01:12 thanks for the cirrus trick 2015-09-17 09:02:46 what do you use for the console, the default vnc? 2015-09-17 09:03:04 no graphic at all, only CLI 2015-09-17 09:03:36 so how do you connect to your guest without networking? 2015-09-17 09:04:07 after initial install, I reboot with networking enabled 2015-09-17 09:04:24 how do you do the initial install? :) 2015-09-17 09:04:44 setup-alpine on the console 2015-09-17 09:04:47 install ssh 2015-09-17 09:04:48 reboot 2015-09-17 09:05:09 next boot is with networking enabled with openvswitch on the host 2015-09-17 09:05:20 so back to the original question: what do you use for the console, the default vnc? :) 2015-09-17 09:05:50 nothing, the console is in plain text 2015-09-17 09:06:46 you get the output of the console in the same terminal you launched qemu? 2015-09-17 09:07:12 it takes control of my ssh session. poweroff after initial install then "screen qemu" (with network). and screen -r to getr access to it 2015-09-17 09:07:29 yes, same terminal 2015-09-17 09:08:24 aha! the -curses option! 2015-09-17 09:08:34 indeed 2015-09-17 09:09:31 if i use that option, i get "1024 x 768 Graphic mode" after the seabios 2015-09-17 09:10:16 that's it! That's why I need to issue "grsec nomodeset" at boot prompt 2015-09-17 09:11:14 on my supermicro boxes, it is enough. Seems that "-vga cirrus" is need in top of that on my poor laptop 2015-09-17 09:11:32 *needed 2015-09-17 09:12:23 I also add the nomodeset option in syslinux at install time. 2015-09-17 09:14:05 hmm, i only see to get my keyboard input echoed once it hit enter 2015-09-17 09:15:41 here, the keyboard echo is fine 2015-09-17 09:16:24 what is your host os? 2015-09-17 09:16:31 AL 3.2.3 2015-09-17 09:17:10 last question :) what do you get from qemu that lxc doesn't give you? 2015-09-17 09:17:48 I want to simulate 7 boxes before going to the Data-Center ;) 2015-09-17 09:18:40 In prod I use both KVM and LXC together 2015-09-17 09:20:25 I tend to group LXC in KVMs. Each KVM is for dedicated customers. LXC are for services. 2015-09-17 09:21:04 when you say KVM, linux-kvm or qemu? 2015-09-17 09:21:15 qemu-kvm 2015-09-17 09:22:29 on baremetal, I run only KVM. in turn, in KVM I run LXC. 2015-09-17 09:22:59 still not with you. this? http://www.linux-kvm.org 2015-09-17 09:23:43 But here a customer wants a setup with several hosts and docker. So I want to prepare it on a lab machine. 2015-09-17 09:24:00 yes kvm.so is my friend ;) 2015-09-17 09:24:17 kvm.ko sorry 2015-09-17 09:24:27 you should definitly give xen a try 2015-09-17 09:24:35 alpine-xen is my friend :) 2015-09-17 09:25:38 I suppose at some stage I will have to give it a try ;) 2015-09-17 09:26:03 you'll never look back 2015-09-17 09:30:24 for now KVM let me launch a VM on a regular host with standard kernel. For Xen, I suppose I need a box prepared for that. 2015-09-17 09:33:59 Jean-Scotch: yes you need to (read should) install alpine-xen on bare metal. it is just as simple and elegant as alpine, as the Dom0 *is* alpine 2015-09-17 09:34:27 but if you already have alpine on bare metal, try setup-xen 2015-09-17 09:34:46 (although i've never used it. always alpine-xen) 2015-09-17 09:36:59 I will when my lab machine arrives (it's on the road as far as the tracking service tells me ;)) 2015-09-17 09:38:57 Also, I still have 6 MSWindows VirtualBoxes running on a Debian Box that I would like to migrate to an Alpine box... maybe this will be the chance for Xen ;) 2015-09-17 10:40:41 any users of s3 storage? how do you consume it? 2015-09-17 11:36:43 cornfeedhobo: re go static: run objdump -p gosu | grep NEEDED 2015-09-17 11:39:46 cornfeedhobo: sorry, did not notice that Shiz helper you :) 2015-09-17 14:32:46 uggedal: yea, i got it solved, but that is still very nice of you to chime in 2015-09-17 14:35:50 ya'll are great. very glad i found out about this project. 2015-09-17 15:01:53 can anyone suggest a minimal PHP Dockerfile using alpine? 2015-09-17 15:37:10 hendry: sorry, no 2015-09-17 15:37:39 you mean a docker container that has php installed? 2015-09-17 15:38:02 you may need to build one. there aren't many out there 2015-09-17 15:42:45 hendry: installing php is trivial, but you'll probably still need to configure it 2015-09-17 15:44:46 i found one i like https://github.com/andrew-d/docker-nginx-php.git 2015-09-17 15:45:04 ACTION is trying to plot something quite simple with rrdtool 2015-09-17 15:54:13 any suggestions for a light system monitoring tool (ongoing+history+graphs)? 2015-09-17 15:56:50 <_ikke_> I'm running zabbix on a raspberry pi 2015-09-17 15:58:57 hendry: that dockerfile's p odd 2015-09-17 16:01:52 observations: you can root your supervision tree on another runsv, which will work better than that wrapper 2015-09-17 16:03:20 runsv will actually do a decent job of reaping zombies so you don't have to implement that in shell 2015-09-17 16:03:38 I did that at first, it was kind of a mess 2015-09-17 16:04:02 it's not that you can't do it, but I wouldn't have trusted it 2015-09-17 16:04:46 the env export doesn't seem necessary, runsvdir doesn't stomp on the environment afaik 2015-09-17 16:06:20 could honestly just apk -X '@testing http://dl-4.alpinelinux.org/alpine/edge/testing' -U add runit@testing, tbh 2015-09-17 16:09:10 the extra level of indirection with /etc/service doesn't seem super-necessary here, just rebuild your container for changes 2015-09-17 16:10:11 I'd check for/add the php group+user in the svdir personally instead of doing it here, but that's not really a significant thing 2015-09-17 16:10:34 can just COPY nginx.conf /etc/nginx/ directly 2015-09-17 16:11:09 the php.ini thing is like, okay, but with the amount of editing going into that php-fpm.conf might as well just COPY that in, too 2015-09-17 16:12:19 honestly if you don't mind your things getting SIGKILLed on container exit you could also just use runsvdir directly as root, but ... well, usually I care about things getting KILLed nanyway 2015-09-17 16:21:10 icicled: sysstat 2015-09-17 16:21:23 it's the old solaris sar ported to linux 2015-09-17 16:22:24 don't expect fancy graphics though, you'll have to roll your own in ksar, librecalc or gnuplot (depending on how brave you feel) 2015-09-17 16:22:46 if you just want network stats, use darkstat 2015-09-17 16:23:18 if you want the full blown thing with a gazillion config options, go for collectd 2015-09-17 16:33:24 icicled: this single box or a bunch or a network? 2015-09-17 16:47:06 just a single box 2015-09-17 16:49:25 uh... anyone knows what happened to libsodium? It's in the aports tree, but not packaged 2015-09-17 16:50:03 it also builds for me 2015-09-17 16:51:13 jomat: are you on stable 2015-09-17 16:51:55 i use pinning: 2015-09-17 16:51:56 @testing http://nl.alpinelinux.org/alpine/edge/testing 2015-09-17 16:52:07 and here it's missing, too: https://pkgs.alpinelinux.org/packages?name=sodium&repo=all&arch=x86_64&maintainer=all 2015-09-17 16:52:48 it's in main 2015-09-17 16:52:57 so @edge 2015-09-17 16:53:12 jomat: you're search exact name 2015-09-17 16:53:24 https://pkgs.alpinelinux.org/packages?name=%sodium% 2015-09-17 16:53:29 oh m( 2015-09-17 16:53:44 Oh gosh… An internal server error occurred. Please try again later. 2015-09-17 16:54:56 yep 2015-09-17 16:54:57 @edge http://nl.alpinelinux.org/alpine/edge/main 2015-09-17 16:55:01 there we are 2015-09-17 16:55:03 thank you :-) 2015-09-17 16:56:11 is node 4+ going up soon on edge? 2015-09-17 16:56:22 for armv6 2015-09-17 17:00:45 atomi, edge/arm builder was offline for a while. it's back up now, but it'll take a while to catch up 2015-09-17 17:01:04 fabled, sounds good 2015-09-17 17:01:06 progress can be followed at build.alpinelinux.org 2015-09-18 04:04:13 anyone ever see this arm box? http://solid-run.com/freescale-imx6-family/cubox-i/cubox-i-specifications/ 4g ram 2015-09-18 06:33:47 uh does busybox has some kind of use-without-initialize in adduser 2015-09-18 06:35:26 https://gist.githubusercontent.com/muhmuhten/ca5a3777f6eacab7a118/raw/5ca472b4ad32f9e28014e146f186f1ec029b4258/- 2015-09-18 06:36:37 ammunta: not adding homedir you mean? 2015-09-18 06:38:13 ammunta: no the weird trash *after* the homedir 2015-09-18 06:38:53 er 2015-09-18 06:38:54 uggedal: 2015-09-18 07:09:24 morning 2015-09-18 07:11:54 ammunta: weird trash? don't follow. alpine don't have a /etc/skel 2015-09-18 07:12:11 e.g. c:x:1002:2:Linux User,,,:/home/c::60:60:mysql:/var/lib/mysq 2015-09-18 07:12:15 the :60:60:mysql:/var/lib/mysq 2015-09-18 07:12:19 should not be there 2015-09-18 07:12:24 that's the shell field 2015-09-18 07:15:27 might be a bug 2015-09-18 07:15:50 that mysql line seems malformed 2015-09-18 10:57:15 ncopa: what is needed to start creating community/ ? 2015-09-18 11:04:29 i suppose a first commit 2015-09-18 11:04:36 then build server config 2015-09-18 11:07:52 i would like to fix the error reporting too first 2015-09-18 11:08:19 that is make buildrepo publish the build progress via hook 2015-09-18 11:15:24 what goes in community/ ? 2015-09-18 11:17:05 one thing more 2015-09-18 11:17:31 i think we want buildrepo uplaod 'main' before continue to build testing 2015-09-18 11:17:33 in case testing fails 2015-09-18 11:17:41 and in case community build fails 2015-09-18 11:17:54 then we dont want prevent that upload a successful build of main 2015-09-18 11:22:39 ScrumpyJack: things which are built for each release (like main) but wont get longterm support 2015-09-19 00:09:59 anyone ever tried building PhantomJS on musl? 2015-09-19 00:10:26 curious if i'm about to venture down a huge rabbit hole... 2015-09-19 00:45:55 https://en.wikipedia.org/wiki/Authbind can we get authbind or such in alpine 2015-09-19 00:46:14 it seems right up our alley 2015-09-19 00:46:32 but the implementation is 2015-09-19 00:46:59 I'm not familiar enough with musl to know whether it would work 2015-09-19 03:46:16 someone running virtualbox on alpine-linux? 2015-09-19 03:46:24 not possible I assume? alternatives? qemu? 2015-09-19 03:50:27 Runs fine for me. 2015-09-19 03:50:35 ^VBox 2015-09-19 03:52:05 Mmwwaaaa, how did you install it? compiled from source? 2015-09-19 03:53:35 No, from ISO 2015-09-19 03:57:39 k 2015-09-19 03:57:47 Mmwwaaaa, ty for the feedback :) 2015-09-19 04:02:14 np 2015-09-19 05:23:59 I want to compile dwm on alpine-linux, but it needs stdio.h 2015-09-19 05:24:11 apparently it's not on the system? 2015-09-19 05:24:51 apk add musl-dev :)) 2015-09-19 05:28:53 you probably actually want build-base 2015-09-19 13:56:48 andyshinn: I got it to build on musl: https://github.com/voidlinux/void-packages/tree/master/srcpkgs/phantomjs 2015-09-19 13:57:36 andyshinn: sadly it did not work very well so I switched to slimerjs 2015-09-19 14:40:48 uggedal: thanks, I think disabling breakpad is the last piece I may need 2015-09-19 14:46:46 andyshinn: keep in mind that I never got it to function properly at run-time 2015-09-19 14:51:40 nice to see that void can contribute back to alpine :) 2015-09-19 15:43:55 anyone using edge branch? 2015-09-19 15:43:59 how "stable" is it? 2015-09-19 15:53:11 myztic: in my experience, it is about as stable as any other software at HEAD, you can occasionally run into broken things and broken packages 2015-09-19 17:34:38 hmm, phantomjs compiles now, but definitely not working 2015-09-19 17:34:46 i wonder if it is because breakpad was removed... 2015-09-19 17:35:19 uggedal: do you remember if you tried the patches from https://code.google.com/p/google-breakpad/issues/detail?id=631? 2015-09-19 18:34:31 hello everyone! 2015-09-19 18:37:58 I know that alpine uses musl and (thanfully)doesn't include systemd, so I was wandering if it's possible to install fglrx 2015-09-19 18:41:33 <_ikke_> Nothing wrong with systemd :P 2015-09-19 18:42:59 I have absolutely no intention of starting another flamewar on the subject. 2015-09-19 18:44:00 systemd doesn't work with musl anyway :) 2015-09-19 20:04:09 chain: i wouldn't rely on it 2015-09-19 20:04:31 they are compiled against glibc, while musl provides a somewhat glibc-compatible API it probably won't be perfect 2015-09-19 20:04:43 [but you can absolutely try] 2015-09-19 20:06:10 Shiz: Thanks! 2015-09-19 20:06:46 s/API/ABI/ 2015-09-19 20:55:49 There's also one other thing that I wanted to mention - do you know if anyone has considered adding Alpine linux support to octopi package manager? 2015-09-19 20:57:49 andyshinn: no, I did not use those patches since the bundles version of breakpad in phantomjs is ancient 2015-09-19 20:58:26 that was the main reason for disabling it. if it were more current it would not be that hard to get compliant with musl 2015-09-19 20:59:03 yea, i gave up... not sure what is different between Alpine and Void that doesn't allow it to work 2015-09-19 20:59:16 does not work on void either 2015-09-19 20:59:22 compiles fine, but the bin just locks up forever waiting for something 2015-09-19 20:59:24 compiles, but does not work runtime 2015-09-19 20:59:31 as I stated earlier 2015-09-19 20:59:34 yep 2015-09-19 20:59:35 i tried a bunch of examples in the void package and it worked 2015-09-19 20:59:42 on glibc? 2015-09-19 21:00:01 works fine on void in glibc, but not on musl 2015-09-19 21:00:04 oh i thought void was all musl... maybe i tried the glibc version, ha 2015-09-19 21:00:22 you probably did :) 2015-09-19 21:00:36 but slimerjs is pretty nice 2015-09-19 21:01:11 i was trying it for another project already using phantom.. i don't really have choice of upstream acceptance test kit :\ 2015-09-19 21:01:25 not headless, but firefox as backend has way more html5-ish feature support than the qt webkit version in phantomhs 2015-09-19 21:01:35 going to scrap the idea for now... phantom is so damn heavy weight, takes nearly an hour just to compile the package from scratch 2015-09-19 21:01:53 slimerjs is 90% api compatible with phantom 2015-09-20 01:26:55 is there a site where you can obtain some infos about the main developers of alpine-linux? 2015-09-20 01:30:20 https://pkgs.alpinelinux.org/packages lists a few maintainers 2015-09-20 01:44:13 I get an error with udhcpc failed to get a dhcp lease, network is down 2015-09-20 01:44:24 Though it isn't 2015-09-20 01:45:30 if I do "udhcpc -i eth0 -q" it just spits out error messages 2015-09-20 07:35:02 myztic not to much info but a starting point http://wiki.alpinelinux.org/wiki/Alpine_Linux:IRC_Availability 2015-09-20 07:35:57 myztic on every rls note there are list of contributors for that specific rls http://git.alpinelinux.org/cgit/aports/log/?h=v3.2.3 2015-09-20 09:45:57 hello, I want to try out alpine linux as a debian replacement on a low-spec laptop, but I don't know which ISO to download to try out 2015-09-20 09:46:01 hello :) 2015-09-20 09:46:09 could someone tell me what the "vanilla" iso is? 2015-09-20 09:49:13 is it 64 or 32 bit? 2015-09-20 09:49:27 32 Bit 2015-09-20 09:50:15 http://wiki.alpinelinux.org/cgi-bin/dl.cgi/v3.2/releases/x86/alpine-3.2.3-x86.iso iirc. 2015-09-20 09:52:17 ok, this is the "standard" iso. I'll try it out, thanks :) 2015-09-20 09:55:35 george_: vanilla is without grsec patches 2015-09-20 09:55:57 mini means its smaller than regular (which is fine if you have inet) 2015-09-20 09:56:50 oh, ok. I need to read about this. I want to try alpine out just because it should be less ressources than debian and cuz it's systemd-free, not because it is more secure... ^^ 2015-09-20 09:59:29 <_ikke_> Note that Alpine has no graphical system installed by default 2015-09-20 10:00:03 that is no problem for me, as long as the installation is not much more complicated than an arch :O 2015-09-20 10:01:37 thanks for your info, i thought that the standard iso has X :D there should be an entry about this in the wiki ;) 2015-09-20 11:39:32 what on earth is it that i've got to do to be able to fsck ext4 from a rescue shell 2015-09-20 11:40:08 why is busybox calling to fsck.auto if that is not there by default 2015-09-20 19:40:03 how tough is it to upgrade alpine linux from version to another, and how stable is edge (compared to maybe debian testing / arch / gentoo) 2015-09-20 19:44:46 How do I get my testing system to update components that were part of main when it was first built? 2015-09-20 19:45:37 I have a stock 3.18.20-1 kernel but testing is at 3.18.21-0 so they don't match up. 2015-09-20 20:07:01 crow, meh, that's when I am sleeping :D 2015-09-20 22:27:12 I guess everyone who uses Alpine lives in European timezones. :) 2015-09-20 22:27:50 :-) 2015-09-20 22:29:19 :P 2015-09-20 22:29:32 jomat: Do you use the testing branch? 2015-09-20 22:30:00 I mix stable, testing and edge 2015-09-20 22:30:38 these are my repos :-) 2015-09-20 22:30:39 http://nl.alpinelinux.org/alpine/v3.2/main 2015-09-20 22:30:39 @edge http://nl.alpinelinux.org/alpine/edge/main 2015-09-20 22:30:39 @jomat /home/jomat/packages/testing/ 2015-09-20 22:30:39 @testing http://nl.alpinelinux.org/alpine/edge/testing 2015-09-20 22:32:20 Looks like I just hosed my test system by rebooting. 2015-09-20 22:32:29 Mounting root: failed. 2015-09-20 22:32:36 initramfs emergency recovery shell 2015-09-20 22:33:05 do you know why it fails? 2015-09-20 22:33:40 I can't see any obvious cause but I'm almost sure it's because I did an install of main then tried to upgrade it to testing. 2015-09-20 22:34:03 can you mount it manually in the initramfs shell? 2015-09-20 22:34:23 So I have a 3.18.20-1 kernel and no idea what in the initramfs and modules. 2015-09-20 22:34:55 hmm 2015-09-20 22:34:58 what fs do you use? I need this kernel command line to boot from ext4: 2015-09-20 22:34:59 fdisk -l shows nothing. 2015-09-20 22:35:02 /vmlinuz-grsec alpine_start=lvm,cryptsetup cryptroot=/dev/root/alpine cryptdm=root-alpine_crypt "root=-text4 /dev/mapper/root-alpine_crypt" ro 2015-09-20 22:35:26 Right now it's just a plain ext4. This is a completely stock xen install. 2015-09-20 22:35:43 Well except that I told it I want testing now too. 2015-09-20 22:36:12 I never used the install scripts… 2015-09-20 22:36:29 dunno how they set up the system 2015-09-20 22:36:43 This is my first experience so I'm not hardcore enough on Alpine to do everything by hand. :) 2015-09-20 22:36:47 fdisk -l shows nothing? 2015-09-20 22:36:55 Yeah nothing at all. 2015-09-20 22:37:00 Probably no device nodes built. 2015-09-20 22:37:00 uh 2015-09-20 22:37:22 Yep no /dev/sd* nodes. 2015-09-20 22:38:23 I'm going to try rebooting and see if maybe there's a bad grub line. 2015-09-20 22:39:03 do you have some live image like http://grml.org/ where you can mount your disk to check if the installed kernel modules fit your kernel version and where you can rebuild your initramfs? 2015-09-20 22:39:39 I could but to be honest I'd rather just nuke the whole thing and reinstall. 2015-09-20 22:39:47 That would probably be faster. 2015-09-20 22:41:11 Also I'm almost certain that my kernel modules don't all fit the kernel version since there appears to be no way to get apk to upgrade the kernel to testing. 2015-09-20 22:41:15 yes 2015-09-20 22:41:33 did you use pinning? 2015-09-20 22:41:44 I don't think so. 2015-09-20 22:42:21 Really all I wanted was to get zfs installed and since that is only in testing I just told it to start using that branch and installed the packages. 2015-09-20 22:42:37 I'd recommend pinning (just adding the @testing tags in your apk/repositories) and just install needed packages: apk add somepkg@testing 2015-09-20 22:42:39 But I assumed it would upgrade the kernel. 2015-09-20 22:43:17 Ah ok… i don't know the current state of zfs+alpine 2015-09-20 22:43:17 Ok, but how does that fix the fact that there's no 3.18.21-0 kernel in testing? 2015-09-20 22:44:25 Also, I think there's a bit of a design flaw in that the initramfs that gets created has no versioning so it looks like it will always overwrite the current one. 2015-09-20 22:44:41 3.18.21-r0 is in edge 2015-09-20 22:44:54 Hmm. 2015-09-20 22:45:04 So what I need to do is enable edge not testing. 2015-09-20 22:45:13 What could go wrong? :) 2015-09-20 22:45:55 :-) 2015-09-20 22:46:36 Ok, I'm giving up on this machine. I don't think there's a point in trying to repair it. 2015-09-20 22:51:36 The installer sure does seem to want to pull a lot of stuff from the network. 2015-09-20 22:52:54 Does it do a network install any time it detects an internet connection? 2015-09-20 22:56:51 Looks like it always requires a network. Hmmm. 2015-09-20 22:57:03 Someone should mention that on the wiki. 2015-09-20 23:20:31 no packages for zfs in edge. :/ 2015-09-20 23:24:19 but on testing. 2015-09-20 23:24:51 So I guess I have to enable every repository and hope that apk doesn't leave my system a complete mess. 2015-09-20 23:25:37 This seems....less than optimal. 2015-09-20 23:34:42 there are no "major" third-party repos, right? 2015-09-20 23:35:06 don't have time to dig into package creation yet, so i figure it's worth asking :) 2015-09-20 23:47:03 Hmm well so far I have a system that appears to work with all 3 repositories enabled and everything upgraded. 2015-09-20 23:49:52 I mean the system boots but I can't imagine it's going to be stable like this. I guess I'll mess around with it and see what happens. 2015-09-21 00:01:54 anyone running alpine-linux on encrypted lvm, I am kind of having trouble with creating the symlink 2015-09-21 00:02:22 http://wiki.alpinelinux.org/wiki/LVM_on_LUKS 2015-09-21 00:02:48 I am talking about: 'apk fix' will give an error for missing modules - fix with a symlink in /lib/modules & rerun 'apk fix' above 2015-09-21 00:21:12 nixdork: 3? 2015-09-21 00:22:10 nixdork: nvm. i see. 2015-09-21 01:01:28 myztic: What symlink are you talking about? 2015-09-21 05:57:05 myztic if you did found the issue, you could submit a bug report. 2015-09-21 06:14:39 nixdork, I don't know from where to what to create the symlink 2015-09-21 06:17:33 apk fix --root /mnt linux-grsec says depmod: can't change directory to '/lib/modules/3.18.20-1grsec': No such file or directory 2015-09-21 06:18:08 Yet I have a folder 3.18.20-1-grsec in /lib/modules and in /mnt/lib/modules 2015-09-21 06:20:26 nixdork, I am supposed to fix the issue with a symlink, but how? 2015-09-21 06:50:59 Are you sure /lib/modules/3.18.20-1grsec is the right path? 2015-09-21 06:51:18 I would think it should be /lib/modules/3.18.20-1-grsec or maybe /lib/modules/3.18.20-r1-grsec 2015-09-21 06:53:47 nixdork, that's a typo 2015-09-21 06:53:56 Ok. 2015-09-21 06:54:46 it says /lib/modules/3.18.20-1-grsec 2015-09-21 06:54:59 I have this folder in /lib/modules and /mnt/lib/modules 2015-09-21 06:56:34 Hmm why do you have a /mnt/lib/modules? 2015-09-21 06:56:42 Are you in the middle of the install? 2015-09-21 06:56:47 yes 2015-09-21 06:56:53 Ah 2015-09-21 06:57:11 Did you have to kill the install and run it again? 2015-09-21 07:01:42 nixdork, my X just crashed an brought down my 2 virtual machines running :/ 2015-09-21 07:01:58 So I guess in a way 'problem solved' as I will have to go through the installation again 2015-09-21 07:02:04 s/an/and 2015-09-21 07:03:43 :( 2015-09-21 07:03:52 If I run into the problem again, I'll post it here, if not and I know what I did wrong, will do so too 2015-09-21 07:07:18 Are you trying to use apk to upgrade anything during the installation? 2015-09-21 07:08:28 I found its best to just get the install done and reboot before doing anything else. 2015-09-21 07:09:06 FUCK HAN UNIFICATION 2015-09-21 07:09:17 wait sorry wrong channel ignore me 2015-09-21 07:09:35 WAT 2015-09-21 07:09:48 nixdork, I pretty much followed the instructions, so in the beginning setting up apkrepots and updating them, and installing some needed software with apk add, like cryptsetup lvm2 and so forth 2015-09-21 07:10:16 but after that no apk use until the very end, where you are supposed to do the apk fix 2015-09-21 07:10:32 s/apkrepots/apkrepos 2015-09-21 07:10:47 Oh hmm. 2015-09-21 07:11:47 nixdork, since I have to go through the whole process again anyway, there is no real need to speculate now 2015-09-21 07:12:09 let's see if I hit the problem again, maybe I don't this time somehow 2015-09-21 07:13:26 I will also save machine state at the very end, so I can try to solve the problem without consequences 2015-09-21 07:13:30 :) 2015-09-21 07:13:52 I found that killing and restarting the installer doesn't work because it marks the hard drive you selected as used and unavailible. 2015-09-21 12:43:44 trying to use those instructions http://wiki.alpinelinux.org/wiki/LVM_on_LUKS 2015-09-21 12:44:12 however I think there is some mistake with using $MNT 2015-09-21 12:46:11 $MNT by default points to /root and mkinitfs -c $MNT/etc/mkinitfs/mkinitfs.conf -b $MNT 2015-09-21 12:46:22 -b does not accept $MNT as an option 2015-09-21 12:55:32 these instructions are really old I see, the author has put updated information on his website regarding this subject 2015-09-21 12:55:34 can be found here http://it-offshore.co.uk/linux/alpine-linux/25-alpine-linux-luks-encrypted-installations 2015-09-21 12:55:45 I will try out these 2015-09-21 13:42:12 myztic - you can : mkinitfs -c /mnt/etc/mkinitfs/mkinitfs.conf -b /mnt 2015-09-21 13:42:54 BitL0G1c, that gave some error message too, though I don't know exactly what it said 2015-09-21 13:42:56 the scripts linked at the the top of the encrypted LUKS page make the whole process much easier 2015-09-21 13:43:31 BitL0G1c, I am using those as we speak 2015-09-21 13:43:45 Hope I will get it done with those 2015-09-21 13:44:19 ok 2015-09-21 13:44:37 BitL0G1c, ty though for getting back to me :) 2015-09-21 13:44:52 this channel is astonishingly inactive for 150+ users 2015-09-21 13:45:06 ;-) np 2015-09-21 14:24:16 myztic: stuff works 2015-09-21 14:24:50 In my case, it doesn't, but I hang out here anyway :P 2015-09-21 14:24:58 hehe ok 2015-09-21 14:25:24 Still hoping to get an Alpine host set up, but for now.. Arch 2015-09-21 14:25:35 ACTION really likes her new setup :D 2015-09-21 14:25:50 Ishiney 2015-09-21 14:39:44 i've been wanting to switch systems to the LBU / boot to ram mode for 2 years 2015-09-21 14:39:53 but other than that, it's all doing nicely 2015-09-21 14:49:08 LBU? 2015-09-21 14:49:32 <_ikke_> Local BackUp? 2015-09-21 15:08:29 BitL0G1c, if I do mkinitfs -c /mnt/etc/mkinitfs/mkinitfs.conf -b /mnt 2015-09-21 15:08:53 it says: /mnt/lib/modules/3.18.20-1-grsec does not exist or is not a directory 2015-09-21 15:13:39 BitL0G1c, I can do apk fix --root /mnt linux-grsec but it then spits out "depmod: can't change directory to 'lib/modules/3.18.21-0-grsec': No such file or directory" and in the instructions it says to fix this with a symlink, though I don't really know what to do tbh 2015-09-21 15:22:21 myztic - cd /lib/modules && ls 2015-09-21 15:22:58 BitL0G1c, shows two folders, one 3.18.20-1-grsec and one called firmware 2015-09-21 15:23:01 ln -s xxxxx 3.18.21-0-grsec 2015-09-21 15:23:25 ln -s 3.18.20-1-grsec 3.18.21-0-grsec 2015-09-21 15:24:00 this happens when the current kernel is ahead of the kernel on the iso 2015-09-21 15:24:39 BitL0G1c, could you expand on what I have to do? 2015-09-21 15:24:49 in which directory do I create what symlinks? 2015-09-21 15:25:51 cd /lib/modules 2015-09-21 15:26:07 ln -s 3.18.20-1-grsec 3.18.21-0-grsec 2015-09-21 15:26:40 BitL0G1c, you mean cd /mnt/lib/modules? 2015-09-21 15:27:03 cd /lib/modules is read-only 2015-09-21 15:27:05 booting from iso can update kernel? 2015-09-21 15:28:38 this happens when the current kernel is ahead of the kernel on the iso 2015-09-21 15:30:53 myztic - i think you need the symlink in /lib/modules (where depmod is complaining) - "depmod: can't change directory to 'lib/modules/3.18.21-0-grsec': No such file or directory" 2015-09-21 15:31:05 I think darkfader is just keeping me in suspense :( 2015-09-21 15:31:28 BitL0G1c, but it complains ln: 3.18-21-0-grsec: Read-only file system 2015-09-21 15:31:32 CandyAngel: o'm sorry i might have missed something 2015-09-21 15:31:38 i'm all stuck in work 2015-09-21 15:31:42 oh 2015-09-21 15:32:05 so, umm, alpine can be installed to a disk like any normal linux 2015-09-21 15:32:21 it can also be installed a little differently, so that your root device is a ramdisk 2015-09-21 15:32:34 and the disk holds only packages / extra config 2015-09-21 15:32:44 on boot, it sucks that up from the disk into memory 2015-09-21 15:33:12 it'll run in ram and every change you make needs to be commited 2015-09-21 15:33:21 it's much like a router operating system then 2015-09-21 15:33:45 meaning, it'll keep going even if some irrelevant piece like the boot disk failed ;) 2015-09-21 15:35:52 BitL0G1c, does that mean it's broken currently? :/ 2015-09-21 15:44:14 myztic - did you try using the 2 scripts ? 2015-09-21 15:50:37 BitL0G1c, I am finished for today with alpine linux 2015-09-21 15:50:46 BitL0G1c, yeah I tried, but ran into some issues too 2015-09-21 15:51:08 I have made a snapshot, maybe tomorrow again 2015-09-21 16:39:03 myztic - I just completed an encrypted LVM installation to a GPT disk scheme using the 2 scripts 2015-09-21 16:39:29 BitL0G1c, are you available later on? 2015-09-21 16:41:17 BitL0G1c, or wait, if you are available now I will do it right now and ask you if I hit a problem / uncertainty 2015-09-21 16:46:08 I've updated the instructions at http://it-offshore.co.uk/linux/alpine-linux/25-alpine-linux-luks-encrypted-installations a little 2015-09-21 16:46:46 the commands before you wget are probably not intuitive 2015-09-21 16:46:59 BitL0G1c, :) those weren't the problem 2015-09-21 16:47:05 I'll tell you when I hit problems 2015-09-21 16:47:10 ok 2015-09-21 16:47:11 :) 2015-09-21 16:47:50 with GPT I had to reboot after first creating the disk scheme 2015-09-21 16:50:07 use mbr :) 2015-09-21 16:50:13 yes 2015-09-21 16:50:56 I think I have never used gpt so far... 2015-09-21 16:50:58 :D 2015-09-21 16:51:12 in all those past years ;) 2015-09-21 16:51:48 Oh, I think my funtoo lvm+luks installation might have had gpt, not sure 2015-09-21 16:54:56 chosen an MBR + lvm disk scheme 2015-09-21 17:10:07 finished & booted it 2015-09-21 17:26:10 BitL0G1c, had a reconnect, if you wrote something, I might not have gotten it 2015-09-21 17:35:14 myztic - only that I finished the MBR install & booted the vm 2015-09-21 17:39:15 BitL0G1c, http://dpaste.com/020T420 what device to pick? 2015-09-21 17:39:16 alpine-root? 2015-09-21 17:46:11 just press 'q' to quit & the partitions will be formatted & mounted 2015-09-21 17:48:08 then run: ./setup-disk -m sys -E 2015-09-21 17:58:17 ERROR: unsatisfiable constrains: 2015-09-21 17:58:27 .setup-apkrepos (missing): 2015-09-21 17:58:39 required by: world[.setup-apkrepos] 2015-09-21 18:04:33 setup-apkrepos though is there, and I of course used it to set up the repos 2015-09-21 18:10:13 BitL0G1c, ^ 2015-09-21 19:15:52 myztic - I think just remove .setup-apkrepos from /etc/apk/world 2015-09-21 19:27:02 BitL0G1c, http://dpaste.com/2QBRJQT :/ 2015-09-21 19:30:36 BitL0G1c, now it works, world file had an empty line on top, therefore the error 2015-09-21 19:35:47 ok 2015-09-21 19:36:08 myztic I have successfully booted into my new installation 2015-09-21 19:36:17 BitL0G1c ty very very much 2015-09-21 19:37:25 np - did you just hit enter when it asked "Select LUKS ROOT Device Mapper:" ? 2015-09-21 19:37:50 maybe I need to prevent blank input 2015-09-21 19:41:34 no 2015-09-21 19:41:36 I typed it in 2015-09-21 19:41:43 ok 2015-09-21 19:41:45 If I remember correctly you have to type it in?! 2015-09-21 19:41:49 otherwise it doesn't work 2015-09-21 19:41:53 yess i thought so 2015-09-21 19:41:57 :) 2015-09-21 19:53:40 BitL0G1c, before I buy one, any good shell scripting (especially bash) books you can recommend? 2015-09-21 19:56:33 https://kat.cr/usearch/bash%20category:books/ 2015-09-21 19:58:51 myztic - https://ebooks-it.org/ too 2015-09-21 20:17:09 BitL0G1c, :) TY 2015-09-21 20:18:01 http://www.amazon.com/New-KornShell-Command-Programming-Language/dp/0131827006 2015-09-21 20:18:18 bash just so closed up in features 20 years later, so this is a recommendation 2015-09-21 20:19:14 darkfader, doesn't it matter that the book is meant for the kornshell? 2015-09-21 20:21:04 the differences are only in advanced stuff - which isn't even described in most bash books 2015-09-21 20:21:30 I don't really get the difference of shells anyways tbh 2015-09-21 20:21:36 not my area of expertise 2015-09-21 20:21:57 why it really matters whether one uses bash, zsh, kornshell, csh, and whatever else exists 2015-09-21 20:22:08 bash is probably the standard now, so... bash it is 2015-09-21 20:22:24 after installing grsec latest iso and rebooting my network is not detected http://sprunge.us/Yejc eth0 ...ip: ioctl 0x8913 failed: No such device . i was able to use iso and install alpine to disk... weird 2015-09-21 20:22:33 you need to know at least one reasonably well to see the differences 2015-09-21 20:22:47 darkfader, well yeah, maybe 2015-09-21 20:23:13 at the simplest level (that is 80's sh style) there's almost no difference 2015-09-21 20:23:43 when while loops have bugs you know you're in bash3 2015-09-21 20:23:55 maybe because of this: * Loading modules ...modprobe: can't change directory to '3.18.20': No such file or directory ? 2015-09-21 21:22:51 myztic: Did you buy your Bash book yet? 2015-09-21 21:23:09 nixdork, nope, but I downloaded a bunch 2015-09-21 21:23:09 http://mywiki.wooledge.org/BashGuide 2015-09-21 21:23:52 nixdork, yours? 2015-09-21 21:24:17 GrayCat's 2015-09-21 21:25:22 Further reading is listed at http://mywiki.wooledge.org/EnglishFrontPage 2015-09-21 21:26:18 GreyCat rather. 2015-09-21 22:51:36 heh, i know him from #hpux 2015-09-21 22:51:54 at least i think so 2015-09-22 02:47:09 <`jpg> I'm trying to install a package from aports but I am hitting "ERROR: x86_64/APKINDEX.tar.gz: UNTRUSTED signature" 2015-09-22 02:47:21 <`jpg> Is this a problem with my abuild key? 2015-09-22 02:52:11 <`jpg> Ok, signing the newly created index file in ~/packages/testing/x86_64/APKINDEX.tar.gz seems to have corrected that. 2015-09-22 02:57:30 <`jpg> Ahh neat, switching to edge + testing avoids that entire mess. 2015-09-22 02:57:52 <`jpg> I still want to get my own repository working but for now this gives me working ZFS setup. 2015-09-22 03:18:03 `jpg: That's the solution I used to get zfs running. 2015-09-22 03:18:27 I had to enable all 3 branches. 2015-09-22 03:19:01 Now I just have to figure out why there's no /dev/disk/ so I can import my pool. 2015-09-22 06:54:09 having trouble subscribing to mailing lists. can someone check if I'm added? 2015-09-22 09:31:42 <`jpg> the /dev/disk symlink stuff is powered by udev rules, alpine may not have the same udev setup you are used to. 2015-09-22 11:06:19 http://www.phoronix.com/scan.php?page=news_item&px=Microsoft-Linux-ACS, ahem cough !! 2015-09-22 11:15:56 vkris: its old news that ms is using linux as well. they used it on there cdn for years 2015-09-22 11:16:41 :) 2015-09-22 12:03:49 yesterday they try to move skype on Microsoft-Linux-ACS ... but skype support for linux are bad :P 2015-09-22 12:04:02 tried* 2015-09-22 12:13:19 <_xs> lulz 2015-09-22 12:27:38 this question is maybe a bit off topic, but my phone is broken and my carrier is offering me a LG G4 ... does anyone of you have experience with that phone? 2015-09-22 12:31:52 nope 2015-09-22 12:51:15 leo-unglaub - http://www.amazon.com/LG-Unlocked-Hexa-Core-Android-Smartphone/dp/B00ZFXYD5K#customerReviews - unlock codes are gbp 2.50 on ebay 2015-09-22 13:35:28 It doesn't have a keyboard 2015-09-22 14:57:19 BitL0G1c: thanks, i will have a look 2015-09-22 15:24:14 Will adding v3.0/testing to the repo list with my current v3.2/main cause any issues? 2015-09-22 15:24:47 bontibon, what are you trying to achieve? 2015-09-22 15:25:07 dalias: trying to install a package that's in 3.0/testing but not 3.2/main 2015-09-22 15:26:06 if it was removed it probably didn't work 2015-09-22 15:26:18 it might have still been made against uclibc 2015-09-22 16:18:14 bontibon: testing should not exist under 3.0 2015-09-22 16:18:19 testing should only exist for edge 2015-09-22 16:19:01 ncopa: how to we remove 3.0/testing and 3.1/testing from the repos? 2015-09-22 16:19:50 bontibon: if you want something in testing, you should be running edge only, or you might run into incompatibilities 2015-09-22 18:17:10 BitL0G1c, did you change setup-paritions script? :D 2015-09-22 18:23:10 not lately 2015-09-22 19:05:39 BitL0G1c, going through installation again and making detailed notes, interested when I am finished? 2015-09-22 19:08:55 yes 2015-09-22 19:27:01 me to :), luks lvm is always interesting 2015-09-22 19:40:24 BitL0G1c, crow will let you know when I am done :) 2015-09-22 21:03:33 BitL0G1c, something I can't figure out here 2015-09-22 21:04:42 When I start ./setup-partitions I do sda, mbr, boot 2015-09-22 21:05:00 yes 2015-09-22 21:06:42 I end up with http://dpaste.com/0ADJMSC then I choose device sda1 to configure, type 'N' 'N' (because I don't want LUKS/LVM), mount point /mnt/boot and select ext2 as filesystem 2015-09-22 21:07:13 you don't need to type anythng 2015-09-22 21:07:24 just 2015-09-22 21:07:25 I know, it's default 2015-09-22 21:07:36 But just to make it clearer 2015-09-22 21:09:48 I then configure /dev/sda2 with LUKS and LVM, no problems there, in the end "format the above partitions" (where it says sda1 100M part --- /mnt/boot ext2) I do yes and it exits fine, but when I do ./setup-disk -m sys -E it says Boot Device '/dev/sda1' is not mounted 2015-09-22 21:09:57 I also can't mount it manually 2015-09-22 21:10:23 If I try, I get mount: special device /dev/sda1 does not exist 2015-09-22 21:11:17 fdisk -l however shows /dev/sda1 Size 100M, type Linux 2015-09-22 21:11:42 End up with the same error every time 2015-09-22 21:13:10 ii had that yesterday with a GPT scheme & had to reboot for a partition to be recognised 2015-09-22 21:13:56 you can probably install parted to make the kernel detect the partiitions 2015-09-22 21:15:29 http://www.cyberciti.biz/tips/re-read-the-partition-table-without-rebooting-linux-system.html 2015-09-22 21:47:13 boy am I frustrated 2015-09-22 21:47:21 I'll never install anything ever again... 2015-09-22 21:47:26 :D 2015-09-22 22:46:25 BitL0G1c, I can't tell why it happens and sometimes doesn't happen 2015-09-22 22:46:41 We can have a competition who can install it faster, I know everything blind by now :D 2015-09-22 23:17:50 myztic - I never figured it out either . I'll add a partprobe before the mounting is done 2015-09-22 23:18:14 BitL0G1c, that might do it 2015-09-22 23:18:28 would be a good exercise to try and find the bug?! :D 2015-09-22 23:18:55 BitL0G1c, when you say you have to reboot, what exactly do you mean by that? 2015-09-22 23:18:58 what did you do? 2015-09-22 23:19:37 I just rebooted the kvm I was installing alpine into 2015-09-22 23:52:59 BitL0G1c, let me know if you changed something because I am screwing around with the scripts 2015-09-22 23:59:07 what is the process to getting a package updated? 2015-09-22 23:59:22 e.g. syslog-ng is out of date now 2015-09-22 23:59:32 and the new version has really neat features 2015-09-23 00:00:06 disclaimer: if there is a wiki about this, i am sorry, but i couldn't find it 2015-09-23 00:02:14 myztic - I won't be touching anything until tomorrow 2015-09-23 00:02:23 BitL0G1c, k 2015-09-23 00:02:40 comfeedhobo - which version ? 2015-09-23 00:05:53 BitL0G1c: https://github.com/balabit/syslog-ng/releases/tag/syslog-ng-3.7.1 2015-09-23 00:06:39 i made an rpm for it, but stupidly left the docker container before moving the srpm to the unionfs 2015-09-23 00:07:27 ok - give me a moment 2015-09-23 00:07:30 the only real changes are some more python deps. they added language bindings, and with it, new tests. 2015-09-23 00:07:33 awesome! thanks 2015-09-23 00:08:07 i remember needing pylint & python-nose 2015-09-23 00:08:53 darn, there were a few more. you'll see the errors when you run tests after compiling. thanks again. 2015-09-23 00:14:44 comfeedhobo http://hastebin.com/cadikefuwi.coffee 2015-09-23 00:25:12 BitL0G1c, fdisk /dev/sda to set up /dev/sda1 100M with bootable flag and /dev/sda2 type '8e' / Linux LVM and then use ./setup-partitions, do sda, m(br), b(oot) type N to not destroy the layout and then Quit the partitioning screen, then just configure as normal 2015-09-23 00:25:48 BitL0G1c, it's reliable, so using fdisk rightly updates the partition table and /dev/sda1 is found correctly 2015-09-23 00:33:50 v 2015-09-23 00:34:17 BitL0G1c, might help to use sgdisk manually exactly as you do in the script, either sgdisk does wrong or you use it wrong in the script?! 2015-09-23 00:37:58 I don't think it's a bug in the script - the partitions are created correctly 2015-09-23 00:40:51 BitL0G1c, bug in sgdisk? 2015-09-23 00:44:37 possibly - I did notice a problem using it manually to list partition codes 2015-09-23 00:46:58 I'll look at it tomorrow & try & break things 2015-09-23 00:49:10 BitL0G1c: thanks. i did some searching but i am not sure how to apply that 2015-09-23 00:50:26 http://patchwork.alpinelinux.org/patch/590/ 2015-09-23 00:50:50 oh, nice. thanks! 2015-09-23 00:50:55 still much to learn 2015-09-23 00:51:14 the easiest way to make a build environment is just to setup an alpine lxc container 2015-09-23 00:51:43 i thought that was the case -- it's what i have been using for GO now 2015-09-23 00:51:59 & run through the steps at http://wiki.alpinelinux.org/wiki/Developer_Documentation - takes about 20 mins or so 2015-09-23 00:52:25 oh, that is where it's hidden. everything i could find about apk was basics. thanks 2015-09-23 00:53:31 when you edit /etc/abuild.conf - I find it useful to change CLEANUP="srcdir pkgdir deps" ----> CLEANUP="deps" 2015-09-23 00:53:51 so you can check what goes into the package 2015-09-23 00:54:44 creating an alpine lxc http://wiki.alpinelinux.org/wiki/LXC#Alpine_Template 2015-09-23 00:57:55 & an example lxc config http://hastebin.com/eyiqeyohob.avrasm 2015-09-23 00:59:09 I use libvirt for the bridge as I also run kvm 2015-09-23 01:00:39 if you don't use libvirt some notes for a nat bridge http://it-offshore.co.uk/linux/debian/61-debian-bridging-wireless-lxc-nat-bridge 2015-09-23 01:02:24 the ovsup / ovsdown scripts are useful for integrating openvswitch & lxc 2015-09-23 08:33:01 is there some way to log to stdout with opensshd? 2015-09-23 09:44:50 boah do i hate grub ... 2015-09-23 09:44:58 grub-install: warning: Couldn't find physical volume `(null)'. Some modules may be missing from core image.. 2015-09-23 09:44:58 Installation finished. No problems reported. 2015-09-23 09:45:03 really grub? 2015-09-23 09:45:17 i am so glad that grub is gone after all servers are alpine ... 2015-09-23 10:47:19 I haven't used GRUB for ageeeeees (pretty much since GRUB2 became a thing) 2015-09-23 10:48:13 Can't you replace GRUB ahead of switching to Alpine? 2015-09-23 10:51:06 CandyAngel: better not 2015-09-23 10:51:19 this is the last server running debian and the hardest to replace 2015-09-23 10:51:40 in order for that one to be replaced it first need rust to compile on alpine linux 2015-09-23 10:52:02 because the postfix remote admin console is written in rust ... for some reason ... dont ask me why my co worker did that 2015-09-23 10:57:13 Ah I see 2015-09-23 11:02:03 "remote admin console" are for girls 2015-09-23 11:02:11 (no offence to girls) ;) 2015-09-23 11:09:08 no comment ... 2015-09-23 11:11:17 OMGthatssooffensive 2015-09-23 11:11:23 ACTION gets triggered 2015-09-23 11:11:33 :P 2015-09-23 11:13:00 Obviously, the only true way to admin something is flipping bits in the CPUs memory using butterflies and needles 2015-09-23 11:28:44 leo-unglaub: rust on musl will be a world of pain 2015-09-23 11:29:01 leo-unglaub: musl as target is supported. but not as host 2015-09-23 11:30:27 uggedal: yeah, i fully agree 2015-09-23 11:30:52 well, maybe someone as brilliant as ncopa will find a way :) 2015-09-23 16:11:19 need glibc for nodejs on rpi 2015-09-23 16:12:46 won't work https://gist.github.com/atomi/f1e243a749a8d357676e 2015-09-23 16:27:59 hey folks, is there a way to get lsb-release on alpine? 2015-09-23 16:28:37 the command is needed by a app I am trying to install 2015-09-23 16:30:38 qua-non: just fake it with a bogus lsb-release 2015-09-23 16:31:00 I guess that could wowrk 2015-09-23 18:22:07 Hello, what is the name for apk manual package? 2015-09-23 18:22:39 hm nvm I think apk --help | less will suffice 2015-09-23 18:48:42 I am trying to find a way to run a script during the initramfs phase -- before the root filesystem has been mounted. I was hoping to do this *without* modifying the 'init' script, but I'm not seeing any plug-points. Am I missing anything? 2015-09-23 19:00:06 I'm trying to compile one program using static linking, but I'm unable to get it to link to musl/libc statically. Any suggestions/pointers? 2015-09-23 19:00:27 I have -Wl,-Bstati,-static,-dn and I still get dynamic 2015-09-23 19:08:19 that must not be the exact option because it has a typo in it 2015-09-23 19:08:27 where are you putting this option in the build? 2015-09-23 19:08:33 and what program are you trying to build? 2015-09-23 19:11:45 scheme9 2015-09-23 19:12:07 yeah I missed c cause I was retyping 2015-09-23 19:13:54 i suspect the build system just isn't using your LDFLAGS like you want it to 2015-09-23 19:14:12 its ccflags 2015-09-23 19:14:18 but -Wl should pass it to linker 2015-09-23 19:14:24 linking options go in LDFLAGS 2015-09-23 19:14:32 not all makefiles pass CFLAGS to the linker 2015-09-23 19:15:26 There is no separate ldflags in that make file nor even separate call to linker 2015-09-23 19:15:33 it's quite simple make file 2015-09-23 19:16:13 Lets try llvm maybe\ 2015-09-23 19:21:54 call to $(CC) without -c or similar is a linking command 2015-09-23 19:22:06 you don't ever invoke 'ld' directly unless you're making a kernel or something 2015-09-23 19:26:13 clang fails to link it... 2015-09-23 19:28:32 look at the output of make and the final command line that links the executable 2015-09-23 19:28:43 that needs to have -static on it. 2015-09-23 19:28:48 if it doesn't, there's your problem 2015-09-23 19:39:20 dalias: thansk man. Big fail on my part, when I said make fiel does not have LDDFLAGS, well vim had highlight of, and I'm used to having it on so missed it 2015-09-23 20:08:08 trying to do a gem install libffi: Invalid configuration `x86_64-alpine-linux-musl': machine `x86_64-alpine-linux' not recognized 2015-09-23 20:09:10 going to try an apk add libffi 2015-09-23 20:09:44 atomi: got the same. installed a libffi(-dev) and that fixed it for me 2015-09-23 20:09:52 ah. too slaow :D 2015-09-23 21:01:44 anyone tried to compile perl www:twitter on alpine? 2015-09-23 21:01:55 it takes days 2015-09-23 21:01:57 haha 2015-09-23 21:22:37 Lemming2, I'm failing at nokogiri now something about libxml2 2015-09-23 21:23:40 https://gist.github.com/atomi/2c1a21126cbb4ea311e8 2015-09-23 21:24:56 :/ maybe edge has a newer libxml2 2015-09-23 21:25:52 hmm libxml2.9.2 weird error 2015-09-23 21:26:21 BitL0G1c, any update on your script / the sgdisk issue? 2015-09-23 21:49:23 atomi: 3.2 has libxml 2.9.2 2015-09-23 21:52:01 atomi: https://github.com/sparklemotion/nokogiri/blob/5a6488a39f805f9b657616fb02ec69ef89927797/ext/nokogiri/extconf.rb#L376 2015-09-23 21:52:40 yeah it can't find the libxml2 libs 2015-09-23 21:57:22 uggedal_, had to use --with-xml2-include=/usr/include/libxml2 2015-09-23 22:04:05 Is a stock alpine install supposed to have working udev by default? 2015-09-23 22:04:19 Or does it maybe just have some stubs? 2015-09-23 22:12:10 Looks like it just has stubs. apk add udev brought in 6 packages. 2015-09-23 22:12:37 still can't build nokogiri :/ https://gist.github.com/atomi/74491daf9ffb25e7132c 2015-09-23 22:14:24 nixdork: stock alpine (alpine-base) only has mdev 2015-09-23 22:14:41 I think going to try with ruby-irb... 2015-09-23 22:14:50 bbiaw 2015-09-23 22:15:10 "Successfully installed nokogiri-1.6.6.2" 2015-09-23 22:15:38 uggedal_, yeah just got it 2015-09-23 22:15:41 uggedal_: Not familiar with that at all. I'm guessing the m stands for minimal? 2015-09-23 22:16:12 here's my dockerfile https://gist.github.com/atomi/b39a9f91f57778c4068c 2015-09-23 22:16:15 I probably need udev to do zfs successfully. Not having proper dev entries is a recipe for misery. 2015-09-23 22:16:29 nixdork: probably, its from busybox. you can install eudev though 2015-09-23 22:17:12 py-pygments is for rdiscount 2015-09-23 22:17:15 :) 2015-09-23 22:19:35 Looks like just installing udev stock isn't enough to create /dev/disk though. 2015-09-23 22:56:20 myztic - I did not have any time today - will look at it tomorrow 2015-09-24 01:18:46 how do I use testing repo? 2015-09-24 01:19:03 oh @ 2015-09-24 01:20:21 --repository works :) 2015-09-24 01:20:53 atomi: Also edit /etc/apk/repositories 2015-09-24 01:21:16 yeah docs are super nice 2015-09-24 01:22:44 I still can't get the apk man page to install. 2015-09-24 01:22:59 I think the apk-doc package was empty. 2015-09-24 01:28:13 BitL0G1c, k, was no request or impatient nagging, just asking out of interest ;) 2015-09-24 01:28:40 nixdork: dont think there is an apk-doc yet... but you can use apk --help 2015-09-24 01:41:02 Yeah apk --help is a little bit...inadequate. :) 2015-09-24 03:05:31 http://www.audi.jp/rice/ 2015-09-24 03:06:00 carbs 2015-09-24 03:07:08 wait wtf did I just send that link to every channel I'm on somehow 2015-09-24 03:07:22 whoops 2015-09-24 03:10:46 <_xs> yum riced car 2015-09-24 03:11:43 rice & distros makes me think of gentoo :) http://fun.irq.dk/funroll-loops.org/ 2015-09-24 03:13:08 <_xs> hahah 2015-09-24 03:13:13 <_xs> this old good post 2015-09-24 03:14:50 <_xs> the time loss optimizing every binary isn't necessarily won in the future 2015-09-24 03:15:43 <_xs> but on the other part, it can be sad to drop what your CPU supports 2015-09-24 03:17:39 the important point is that a very small subset of software has anything to do with system performance 2015-09-24 03:17:54 there's no reason to compile coreutils with -funroll-all-loops :) 2015-09-24 03:20:46 <_xs> hmm 2015-09-24 03:21:09 <_xs> i see, it really depands on what's the software supports 2015-09-24 03:24:32 awesome :) https://github.com/atomi/dockerfiles/commit/6720a4e4649ae85b7b66e4c61ff1791c7263d0f4 2015-09-24 03:26:09 can I clean that up even more? 2015-09-24 03:32:33 <_xs> I don't use docker, ruby or jekyll for my part :p 2015-09-24 03:33:01 yeah me too. I was just doing this for fun 2015-09-24 03:35:49 <_xs> haha, then you has avoided spending 600M for th 2015-09-24 03:35:55 <_xs> this stack ? 2015-09-24 03:36:20 yeah 2015-09-24 03:37:47 <_xs> what was it so heavy ? 2015-09-24 03:38:15 the other base image was based on a full install of debian jessie I believe 2015-09-24 03:38:36 <_xs> haa, sure :P 2015-09-24 03:38:36 official ruby docker images are based off of debian 2015-09-24 03:39:33 <_xs> we won't switch to debian 8 at my company 2015-09-24 03:40:03 <_xs> i'm running two alpine servers, all goes right actually 2015-09-24 03:41:33 <_xs> I think we couldn't switch to alpine because the dev's requirements won't fit with musl :/ 2015-09-24 03:43:25 yeah 2015-09-24 03:44:34 atomi: any particular reason against using apk for ... well just nokogiri but 2015-09-24 03:46:04 ammunta, ? there's an apk for nokogiri? 2015-09-24 03:46:29 oh yeah nice 2015-09-24 03:46:49 ammunta, haha I hand't even checked 2015-09-24 03:47:00 *hadn't 2015-09-24 03:50:41 not a lot of traffic on the forums :/ 2015-09-24 03:50:50 http://forum.alpinelinux.org/forum/general-discussion/where-can-we-request-packages-or-learn-about-addingmaintaining-pkgs 2015-09-24 03:50:58 ammunta, thanks 2015-09-24 03:56:56 atomi: it's not that difficult tbh to package ruby gems 2015-09-24 03:57:24 these are perl packages 2015-09-24 03:57:53 they take so long to build on armv6l ... 2015-09-24 06:36:45 <_ikke_> I'm trying to understand a message in /var/log/messages: grsec: From : failed fork with errno ENOMEM by /path/to/process.... 2015-09-24 06:37:24 <_ikke_> I'm ssh'd into that machine with my workstation, but I wonder what it means that it shows my workstation IP there. Anyone a clue? 2015-09-24 07:00:12 _ikke_: I tried ssh with a bad password but i didn't get any grsec notices 2015-09-24 07:01:04 <_ikke_> One process died due to OOM, though, monitoring shows there was 1GB free on the machina 2015-09-24 07:01:06 <_ikke_> machine* 2015-09-24 07:01:23 <_ikke_> It's the path to that process that's in the logs 2015-09-24 07:18:37 <_ikke_> What's the difference between available memory and free memory? (agent values) 2015-09-24 07:18:43 <_ikke_> sorry, wrong window 2015-09-24 07:23:59 can we see the log? 2015-09-24 11:06:06 http://www.alphr.com/google/1001573/i-put-a-spell-on-url-crash-google-chrome-with-16-characters 2015-09-24 11:41:30 yup, that crashed chromium on alpine :) 2015-09-24 11:50:23 ncopa: What is the thing that needs to be done before we can have thunderbird? I wonder if it is something I can help with? I am not a c++ programmer, so if it requires quite a bit of changes to the source I am not to much help. If it is on other levels, I would want to help if I am able. 2015-09-24 11:54:18 we need an APKBUILD for it 2015-09-24 11:54:20 a build script 2015-09-24 11:54:47 my guess is that we could start from the firefox APKBUILD 2015-09-24 11:54:52 might be we need some of the patches there too 2015-09-24 11:55:12 Where do I find the APKBUILD scripts? 2015-09-24 12:14:12 http://git.alpinelinux.org/cgit/aports/tree/testing/firefox 2015-09-24 12:14:30 Thanks 2015-09-24 12:24:22 ncopa: Is there any tool for automatically adding all the md5sums to the end of the APKBUILD file? Or do I need to do it one at the time? 2015-09-24 12:24:29 manually 2015-09-24 12:33:28 nmeum: you use opensmtpd right? 2015-09-24 12:35:27 ScrumpyJack: yeah, why? 2015-09-24 12:35:54 have you implemented any black/grey/whitelisting? 2015-09-24 12:35:57 engblom: abuild checksum 2015-09-24 12:36:40 ScrumpyJack: yes, I use spamd under OpenBSD on alpine I only run OpenSMTPD locally 2015-09-24 12:37:26 yes spamd looks good, but there isn't a linux port is there? 2015-09-24 12:38:06 I don't think so 2015-09-24 12:41:50 https://www.milter.org/ is no more :( 2015-09-24 12:48:31 is there a dmraid tool/driver on alpine 2015-09-24 12:48:42 trying to copy some stupid intel raid 2015-09-24 12:48:55 apk search dmraid has nothing 2015-09-24 13:02:43 engblom: abuild checksum 2015-09-24 13:21:35 darkfader: probably not 2015-09-24 13:23:36 ncopa: are you backporting 4.1 testing grsec patches to 3.18? 2015-09-24 13:25:51 kocka: we did port some of the changes in 4.1 to 3.18 iirc 2015-09-24 13:26:01 i dont know if we will continue with that now 2015-09-24 13:26:17 it was kinda nice to have the 3.14 stable patch as reference 2015-09-24 13:27:47 ncopa: thanks 2015-09-24 13:28:14 i'm copying the two disks using alpine now, it's a good thing to have a sub-raid-level backup 2015-09-24 13:28:25 and then i'll boot some ugly other os 2015-09-24 13:28:56 it seems recent mdadm can lift most of the "isw raid" weight but it's too risky 2015-09-24 13:29:12 oh it can? 2015-09-24 13:29:17 i didnt know 2015-09-24 13:29:39 the fakeride is just stupid imho 2015-09-24 13:29:55 something like mdadm -I -e imsm /dev/md127 2015-09-24 13:29:59 of course it is 2015-09-24 13:30:10 the only reason it exist is that softraid was not supported in some windows xp 2015-09-24 13:30:13 acer was kind enough to sell laptops that ship with a ssd raid0 2015-09-24 13:30:44 so if one disk fails you lose data on both 2015-09-24 13:30:45 :) 2015-09-24 13:31:37 yeah and if the laptop breaks you can't access the data 2015-09-24 13:32:25 you cannot disable it? 2015-09-24 13:32:34 and access the disks indivitually? 2015-09-24 13:42:04 its a raid0(!) 2015-09-24 13:43:18 if they driver / tools are missing as it's in alpine i can only access the disks indiv. but it's not doing anything useful - except for backup, what i'm doing now 2015-09-24 15:53:37 hmm 2015-09-24 15:53:39 https://www.openprinting.org/printer/HP/HP-LaserJet_5M 2015-09-24 15:53:49 i installed hplip@testing but can't seem to see my printer driver 2015-09-24 15:54:24 iirc my printer driver is in gutenprint but we don't seem to have that 2015-09-24 15:58:33 https://packages.debian.org/jessie/printer-driver-gutenprint 2015-09-24 21:55:58 ncopa, are you aware fakeroot does not work on alpine? 2015-09-24 21:56:14 i know how to fix it but not how to make the fix acceptable for upstream 2015-09-24 22:02:30 https://alioth.debian.org/plugins/scmgit/cgi-bin/gitweb.cgi?p=fakeroot/fakeroot.git;a=blob;f=configure.ac;h=6cf46ed9fd951278d546c460169e3580aff7a6c9;hb=HEAD#l304 2015-09-24 22:02:33 - for WRAPPED in __${PRE}x${FUNC} _${PRE}x${FUNC} __${PRE}${FUNC}13 ${PRE}${FUNC}; do 2015-09-24 22:02:37 + for WRAPPED in ${PRE}${FUNC}; do 2015-09-24 22:14:23 dalias: would probably be better to adapt the build system to build in a USER_NS with something like your super_chroot 2015-09-24 22:15:51 yes but it would be nice to fix fakeroot and get the fix upstream 2015-09-24 22:16:03 especially since buildroot uses it and isn't interested in changing afaik 2015-09-24 22:16:14 and fetch distfiles before entering clone with CLONE_NEWNET 2015-09-24 22:16:22 and i think i know the correct way to make the above conditional 2015-09-24 22:16:45 yeah, archlinux uses it still afaik 2015-09-24 22:16:55 #ifdef _STAT_VER 2015-09-24 22:23:44 alternatively it could check for decls (AC_CHECK_DECLS) rather than symbols (AC_CHECK_FUNCS) 2015-09-25 06:57:07 morning. 2015-09-25 08:45:17 any qemu users passing through nic cards to their guests? 2015-09-25 09:18:11 ok done 2015-09-25 09:19:28 CandyAngel: did you get your graphic card passthrough sorted? 2015-09-25 09:19:38 i'm trying to do something similar 2015-09-25 09:26:35 ScrumpyJack: Yeah, but not in Alpine 2015-09-25 09:27:11 Though it should be possible to do what I have in Alpine (providing you can ACS patch the kernel) 2015-09-25 09:28:22 which is not required unless your IOMMU groups are wrong 2015-09-25 10:32:38 ScrumpyJack: i tried to pass through audio card to qemu guest 2015-09-25 10:35:17 you mean direct passthrough vs. virtio? 2015-09-25 10:37:56 wow... was not aware qemu could do that 2015-09-25 10:37:57 direct pass 2015-09-25 10:37:58 vmware beware 2015-09-25 10:39:56 if you succeeded, you may want to document it :D 2015-09-25 10:42:56 i managed to passthru a usb ethernet adapter to my osx vm, now i want to try a gfx card 2015-09-25 10:44:02 although i read that spce and qlx might be a good option 2015-09-25 10:44:12 qxl even 2015-09-25 11:06:20 I use QXL, but not SPICE, so I don't know if it works properly 2015-09-25 11:06:42 Any resolutions much more than 1280x720 are laggy without SPICE 2015-09-25 11:07:15 Which is fine for me, all my non-main VMs don't need much screen estate or have no X 2015-09-25 11:07:30 And my main VMs are GPUpassthrough'd 2015-09-25 11:08:42 CandyAngel: syntax for GPU passthru? 2015-09-25 11:08:56 do you need to use the stub thingy? 2015-09-25 11:09:05 Yeah 2015-09-25 11:09:59 strangly, qemu handles removing a usb device from the host for passthru 2015-09-25 11:10:05 i don't have to do it 2015-09-25 11:11:12 CandyAngel: did you passthou the whole gfx card or can you passthrou just one "port" in a multi port gfx card 2015-09-25 11:11:29 Yeah, using "in-use" USB devices works fine (I passthrough my host keyboard/mouse) 2015-09-25 11:11:35 It's the whole PCIe device 2015-09-25 11:15:12 I can't show you my VM launch script right now.. can't access the host remotely :P 2015-09-25 11:15:29 ACTION *blink* 2015-09-25 11:17:59 Secure host is secure? :P 2015-09-25 11:30:01 ScrumpyJack: I'll be going home soon so I will pastebin my linux-gpu script 2015-09-25 11:35:44 cool 2015-09-25 11:36:33 i've just found a spare agp gfx card, and i seem to have 2 agp ports, so i'll try passthru of the whole device 2015-09-25 11:40:17 I.. don't think that will work 2015-09-25 11:40:46 So let me know if it does :P 2015-09-25 12:05:00 linux can see both gfx cards 2015-09-25 12:10:16 but as soon as i unbind the pci device i want to passthru, kernel hangs 2015-09-25 12:28:03 do i need to pass some options to the kernel at boot to avoid that? 2015-09-25 13:28:43 You're not supposed to unbind it, it has to be bound to either pci-stub of vfio-pci and just passed straight to the VM like that 2015-09-25 13:28:51 s/of/or/ 2015-09-25 13:29:17 ok 2015-09-25 13:29:31 both agp cards are nvidia cards 2015-09-25 13:29:34 is that ok? 2015-09-25 13:30:02 i need nouveau to load for the agp host's agp card 2015-09-25 13:30:23 but i remember you had to blacklist nouveau. is that right? 2015-09-25 13:32:35 No, you can bind specific vendor:device ids with pci-stub so you don't have to blacklist nouveau 2015-09-25 13:32:43 cool 2015-09-25 13:33:14 can i see your the bind if your script? 2015-09-25 13:34:00 just echo "dev" sys/bus/pci/drivers/pci-stub/new_i 2015-09-25 13:34:16 or do i need more 2015-09-25 13:35:17 also, once bind is done, is passing -device pci-assign,host=$dev to qemu enough? 2015-09-25 13:36:38 i see this everywhere in blogs: "echo $dev > /sys/bus/pci/devices/$dev/driver/unbind" 2015-09-25 13:36:41 I don't bind in my script, I bind using vfio-pci and modprobe.d 2015-09-25 13:36:52 are you suggesting that's not needed? 2015-09-25 13:37:15 can you be more specific? 2015-09-25 13:37:58 When my computer boots, the vfio-pci module is loaded and reads a list of pci-ids from /etc/modprobe/vfio.conf to bind against 2015-09-25 13:38:08 This prevents the nouveau module binding to it 2015-09-25 13:38:15 binding to those devices 2015-09-25 13:38:38 You have to do it with pci-stub in Alpine though 2015-09-25 13:38:48 so does vfio-pci have to load before nouveau? 2015-09-25 13:38:49 Because the vfio-pci module is too old 2015-09-25 13:38:52 Yeah 2015-09-25 13:39:03 http://wiki.alpinelinux.org/wiki/User:CandyAngel/KVM_Host_Tutorial 2015-09-25 13:39:24 That's what I had written up so far 2015-09-25 13:39:38 Which doesn't include the pci-stub stuff.. bleh 2015-09-25 13:39:41 Must ahve not saved it 2015-09-25 13:39:56 history? :) 2015-09-25 13:40:57 I was typing it up on my laptop while I was setting it up, but I don't think I saved it (because I didn't get it working) 2015-09-25 13:41:38 i've seen people add pci-stub.ids= to kernel boot params 2015-09-25 13:41:44 does that ring a bell? 2015-09-25 13:42:00 Yeah, that's what I used in ALpine to do it 2015-09-25 13:42:23 I had to write a features file to include the pci-stub module in initram 2015-09-25 13:42:34 Otherwise, it got loaded too late 2015-09-25 13:43:29 http://codepad.org/Smhidcoh 2015-09-25 13:43:37 That's my 3D accelerated Linux VM launch script 2015-09-25 13:44:17 But that probably won't help you.. I doubt any AGP cards have an EFI ROM to be booting using OVMF 2015-09-25 13:44:24 booted* 2015-09-25 13:44:41 I had to buy a GTX970 to get this setup working (GTX570 has no EFI) 2015-09-25 13:46:34 i have 2015-09-25 13:46:39 04:00.0 VGA compatible controller: NVIDIA Corporation GT218 [GeForce 210] (rev a2) 2015-09-25 13:46:42 and 2015-09-25 13:46:48 05:00.0 VGA compatible controller: NVIDIA Corporation GK107 [NVS 510] (rev a1) 2015-09-25 13:47:03 the first is the one i'm trying to passthru 2015-09-25 13:47:19 right, reboot needed 2015-09-25 13:50:08 You may be able to get an EFI ROM for the GT218 2015-09-25 13:52:10 No IOMMU found 2015-09-25 13:52:31 why would i need a ROM for the card? 2015-09-25 13:52:38 Do you not have VT-d support? 2015-09-25 13:52:42 yup 2015-09-25 13:53:03 of course - qemu is running all sorts already 2015-09-25 13:53:19 QEMU doesn't require VT-d 2015-09-25 13:53:30 GPU passthrough does 2015-09-25 13:53:44 ah right 2015-09-25 13:56:01 i have VT-x 2015-09-25 13:56:09 That's why I had to buy a new CPU 2015-09-25 13:56:18 I had the i5-2500K and that doesn't have VT-d 2015-09-25 13:56:37 Intel(R) Xeon(R) CPU E5-1620 0 @ 3.60GHz 2015-09-25 13:56:51 be aware, not only a capable CPU is needed, a capable south/north bridge is also needed 2015-09-25 13:57:17 Yeah, you need a motherboard that does VT-d too 2015-09-25 13:57:21 http://tinyurl.com/8efebw7 2015-09-25 13:57:31 says i have VT-d and VT-x 2015-09-25 13:57:46 It also needs to be enabled in the BIOS/UEFI thing 2015-09-25 13:58:05 let me check /proc 2015-09-25 13:59:17 ScrumpyJack: you will probably have to use inteltool from coreboot in order to know which chipset you have. 2015-09-25 13:59:55 ScrumpyJack: I have one laptop next to me that has a VTd capable CPU, but the chipset doesn't support IOMMU, so it's useless 2015-09-25 14:01:18 perhaps https://en.wikipedia.org/wiki/List_of_IOMMU-supporting_hardware 2015-09-25 14:04:21 you still need to figure out which chipset you have, each CPU can be paired with a couple of chipsets usually (one cheap, one expensive) 2015-09-25 14:05:26 ACTION opens the pc case and stares at the motherboard 2015-09-25 14:06:21 inteltools or dmidecode is easier IMHO 2015-09-25 14:07:20 ok, got dmidecode 2015-09-25 14:07:45 what am i looking for in the output? 2015-09-25 14:08:47 seems like all possible chipsets for Xeon(R) CPU E5-1620 should support VTd 2015-09-25 14:09:28 do you see any C60* item in the list. I've only used inteltool myself, not dmidecode 2015-09-25 14:09:33 just paste the output somewhere 2015-09-25 14:09:49 http://sprunge.us/SCcU 2015-09-25 14:14:51 i'll try passing intel_iommu=on to kernel 2015-09-25 14:16:17 I don't see the chipset version listed there, I'm sorry 2015-09-25 14:16:39 did you check for newer bios versions? 2015-09-25 14:24:42 ok, intel_iommu=on worked 2015-09-25 14:25:29 then probably you have an IOMMU with erratas, and it's disabled by Linux by default? 2015-09-25 14:25:35 i see IOMMU in dmesg 2015-09-25 14:25:49 but now my device isn't grabbed by stub any more 2015-09-25 14:33:46 does anyone know of documentation for running cronjob on alpine linux? is it the same as cron and crontab in ubuntu? 2015-09-25 14:42:43 im out of ideas. i must have just missed something 2015-09-25 15:14:40 ScrumpyJack: I'm sorry, I cannot provide much help with KVM. 2015-09-25 15:19:56 Hi. I'm new to alpine, and have a question about diskless installs. I set up lbu and apk cache, and it's great, however upon reboot it seems packages in cache are not automatically added, despite being listed in /etc/apk/world (specifically privoxy). Am I doing it wrong? How should I add packages to a diskless install? Thanks! 2015-09-25 15:52:08 dc181: if the packages are not on the initial boot media, I have to issue "apk update && apk fix" at the end of the boot sequence. I do it with a /etc/local.d/system.start file and "rc-update add local" 2015-09-25 15:54:27 Jean-Scotch: Thanks, I'll do that. Much easier than making a custom ISO. 2015-09-25 17:25:49 how to disable creating of keys on sshd reload / restart if they are deleted manually from /etc/ssh/ . ssh-keygen: generating new host keys: RSA1 DSA ECDSA 2015-09-25 17:31:03 i see in init.d/sshd the "if ! yesno "${SSHD_DISABLE_KEYGEN}"; then" but where is this option disabled 2015-09-25 17:34:40 ok found it, adding this "SSHD_DISABLE_KEYGEN=yes" to /etc/conf.d/sshd would make init.d happy and no creating the not secure keys 2015-09-25 17:59:27 re-posting my question: does anyone know of documentation for running cronjob on alpine linux? is it the same as cron and crontab in ubuntu? 2015-09-25 18:02:36 tridnguyen: depends on which cron deamon you're using 2015-09-25 18:03:07 uggedal: oh? could you explain a bit more? sorry I'm rather new to cron... not sure what I could do 2015-09-25 18:04:01 well, do you have peridodic jobs you need to run? 2015-09-25 18:04:16 if yes, then cron is the answer 2015-09-25 18:04:16 yes 2015-09-25 18:05:15 well, busybox comes with a crond, started as /etc/init.d/cron 2015-09-25 18:05:35 if I wanna add my own cron job, can I add it to /etc/cron.d? 2015-09-25 18:05:56 tridnguyen: not with the busybox one 2015-09-25 18:06:03 as root, run crontab -l 2015-09-25 18:06:29 then you see that you can add executable scripts in /etc/periodic/* 2015-09-25 18:06:46 if you need a custom time format, use crontab -e and add your specific job 2015-09-25 18:07:06 is there a way to specify the custom job via a script file? 2015-09-25 18:07:29 yeah 2015-09-25 18:08:00 if the intervals of /etc/periodic suits you, just add a executable script to one of its directories 2015-09-25 18:08:22 the logrotate package for instance installs this: 2015-09-25 18:08:25 find /etc/periodic/ -type f 2015-09-25 18:08:27 /etc/periodic/daily/logrotate 2015-09-25 18:09:13 if you need more granularity, use crontab -e, type in your cron time format and the executable you wan't to run 2015-09-25 18:09:20 the default periodic are `15min`, `hourly`, `daily`, `weekly` and `monthly` right? 2015-09-25 18:09:25 yep 2015-09-25 18:10:02 so my question was - is there a way to do something like crontab -e, but through a script file instead? 2015-09-25 18:10:28 I read somewhere that in ubuntu, you can add that script file that contains the time format and executable to /etc/cron.d, and then run `cron` command 2015-09-25 18:10:53 there is, but it's not adviced to do it directly 2015-09-25 18:11:14 oh really? 2015-09-25 18:11:18 cat /var/spool/cron/crontabs/root 2015-09-25 18:11:26 that is what crontab -e as root is editing 2015-09-25 18:11:50 well, the reason I wanted to do that is because I wanted to build a docker-compose file that would set up a bunch of different things, one of them being a docker container that would run these custom jobs 2015-09-25 18:11:51 but the format is quite strict, so if you edit it directly it could fail to parse 2015-09-25 18:12:00 ah I see 2015-09-25 18:12:04 hence crontab -e which validates your input before saving it 2015-09-25 18:12:13 got cha 2015-09-25 18:12:41 but if you're scripting it, editing the file directly is ok 2015-09-25 18:13:25 (after you've validated it with crontab -e) for intsance 2015-09-25 18:13:48 makes sense 2015-09-25 18:13:59 you could also do: 2015-09-25 18:14:12 echo '* * ........' | crontab - 2015-09-25 18:14:19 oh 2015-09-25 18:29:21 hi. this is embarrassing, and i'm not sure if i'm using the right words. i want to to verify the gpg signature for the iso. but all the info i find on how to do that involves a keyserver, and i'm not sure how or where to find one 2015-09-25 18:33:50 tigrmesh: ncopa's gpg pub key is here: https://www.alpinelinux.org/keys/ncopa.asc 2015-09-25 18:34:25 tigrmesh: keyservers are just for getting keys 2015-09-25 18:34:31 just skip past those sections 2015-09-25 18:34:52 https://www.gnupg.org/gph/en/manual/x135.html anyway 2015-09-25 18:49:52 i am lost. is this even remotely close? $ gpg --output alpine-3.2.3-x86.iso.asc --detach-sig ncopa.asc 2015-09-25 18:59:10 what am i missing? 2015-09-25 19:06:42 i get intimidated, then my mind closes, and i feel stupid. and it's all downhill from there. some further hint would be appreciated 2015-09-25 19:12:06 you should read the gnupg manual instead of blindly trying different commandline options 2015-09-25 19:20:41 fair enough 2015-09-25 19:20:49 just a thought: skip checking the validity of the iso 2015-09-25 19:22:16 i've had that thought many times in the last while :) 2015-09-25 19:22:32 i thought i'd do something different 2015-09-25 19:27:08 if you don't understand how gpg works and how can you trust it then? And how did you verify ncopa's key? 2015-09-25 19:29:41 sha512sum alpine-3.2.3-x86.iso 2015-09-25 19:29:43 i didn't. i trust none of it. that's why i'm here 2015-09-25 19:29:44 4ee580eb1a80fb6f72811a6cdb5ec7483ce9aa9f0da8588cf09d1db3ce2a64876e6c085da6d1bbf547e006eab7001d1b79dca72c608ac98c8f4a95cc1281fcf0 alpine-3.2.3-x86.iso 2015-09-25 19:29:51 from nl.alpine.org a few seconds ago 2015-09-25 19:30:09 you'd have to trust nl.alpine.org and me though 2015-09-25 19:30:42 and freenode 2015-09-25 19:30:46 and your isp 2015-09-25 19:30:58 but whatever 2015-09-25 19:31:29 you can check who signed ncopa's key 2015-09-25 19:31:37 and try to establish a trustpath to it 2015-09-25 19:32:41 i checked the sha1 and sha512 numbers earlier. 2015-09-25 19:33:57 if anyone said anything after this, i didn't see it from nl.alpine.org a few seconds ago 2015-09-25 19:34:30 tigr: 19:30 uggedal you'd have to trust nl.alpine.org and me though 2015-09-25 19:34:33 19:30 uggedal and freenode 2015-09-25 19:34:35 19:30 uggedal and your isp 2015-09-25 19:34:37 19:30 uggedal but whatever 2015-09-25 19:39:22 that is all true. 2015-09-25 19:40:28 i trust that the file download was complete 2015-09-25 19:45:17 uggedal, thank you 2015-09-25 19:48:23 uhmm is there some glibc based alpine version ? 2015-09-25 19:48:30 or some guide how to build one ? 2015-09-25 19:51:23 nmeum, thank you. in thinking about your question, i remembered that i have done this before. now i need to find how. 2015-09-25 19:52:31 livcd: no and no 2015-09-25 19:53:08 but well you can still install glibc on alpine and manually link all packages against it 2015-09-25 19:59:35 <_ikke_> 3ttdfksdfsdfkj 2015-09-25 19:59:45 <_ikke_> cvxcvxcvxcvcvsdfsdfdfsfdffdsfdsdfjsdfjsdfkjsdfjsdfjsdfkjsdfjksdfjkdfsdf 2015-09-25 19:59:52 <_ikke_> Sorry :-/ 2015-09-25 20:16:11 _ikke_: been cleaning your keyboard ? 2015-09-25 20:17:48 <_ikke_> No, flaky internet connection 2015-09-25 20:18:00 <_ikke_> running weechat over ssh 2015-09-25 20:20:36 ~. 2015-09-25 20:22:25 <_ikke_> yeah, that caused these messages :P 2015-09-25 20:22:36 <_ikke_> at least, after I tried to see if something responded 2015-09-26 01:01:03 thank you all. i got it sorted 2015-09-26 03:30:27 hi! I am trying to use alpine as base for a docker image containing ruby tools like compass, is anyone doing something like this? I am having problems with libffi at install time... 2015-09-26 08:09:33 <_xs> docker sucks 2015-09-26 11:26:17 Hello everyone 2015-09-26 11:28:17 I have a problem with Xorg, in that I either have no input at all, or X sort of "freezes", not sure which (hard to tell when I have no input). 2015-09-26 11:28:47 I can ssh into the machine, and killing the X process returns me to the console thankfully, and everything works normally at the console. 2015-09-26 11:29:00 <_xs> hello flux_control 2015-09-26 11:29:02 <_xs> dbus enabled ? 2015-09-26 11:29:27 afaik it is 2015-09-26 11:29:45 Though I wouldn't think that's needed for keyboard input to function. Is it? 2015-09-26 11:30:11 <_xs> there's actually a bug with openrc's cache on some services like dbus 2015-09-26 11:30:17 <_xs> maybe it's that 2015-09-26 11:30:33 <_xs> look at the dbus status 2015-09-26 11:30:39 I have a few "errors" in my X log, but they are only for missing video drivers that don't affect me (it tries things like vesa etc., but works with intel, so not really "errors"). 2015-09-26 11:31:04 Well, this isn't a "service" that I'm having a problem with, it's the X input itself. 2015-09-26 11:31:06 <_xs> try running the alpine setup xorg script 2015-09-26 11:31:13 <_xs> ... 2015-09-26 11:31:21 <_xs> dbus is a service 2015-09-26 11:31:25 I can't kill the X server via ctrl-alt-backspace even 2015-09-26 11:31:31 <_xs> and he's giving input to X 2015-09-26 11:31:31 Yes, dbus is. X input isn't. 2015-09-26 11:32:13 Well, dbus is running, and I don't see anything in the logs for problems with it. 2015-09-26 11:32:21 Is there somewhere else I should be checking? 2015-09-26 11:33:06 You mentioned about a bug with a cache. Is there a cache I should remove or something? 2015-09-26 11:34:22 <_xs> there's a script in alpine which manually starting a dbus service 2015-09-26 11:34:30 <_xs> but you might running it by hand 2015-09-26 11:34:38 <_xs> then go to /lib/rc/cache 2015-09-26 11:34:44 <_xs> rm * 2015-09-26 11:34:58 <_xs> and rm /run/openrc/deptree 2015-09-26 11:35:12 <_xs> finally regenerate the cache: /lib/rc/bin/rc-depend 2015-09-26 11:35:41 I'm not running dbus by hand 2015-09-26 11:35:47 <_xs> ... 2015-09-26 11:36:01 <_xs> yes but the alpine Xorg setup script does 2015-09-26 11:36:13 <_xs> and again, there's a problem with rc's cache dependencies* 2015-09-26 11:36:19 <_xs> in this release 2015-09-26 11:36:36 <_xs> that's exactly how I fixed what you describe 2015-09-26 11:36:58 I wasn't complaining, just letting you know since you said I might be running it by hand 2015-09-26 11:37:04 Was just giving info 2015-09-26 11:37:45 <_xs> normally the script add dbus to the default group or something 2015-09-26 11:38:03 Yeah, that's what I have 2015-09-26 11:38:06 <_xs> but it doesn't work, i don't know why 2015-09-26 11:38:09 dbus autostarts on boot 2015-09-26 11:38:24 <_xs> so I saw the same 2015-09-26 11:38:43 <_xs> then after regenerating rc dependencies, it worked and X got inputs working 2015-09-26 11:38:47 <_xs> no freezing anymore 2015-09-26 11:39:09 Is there some way I should be calling rc-depend? 2015-09-26 11:39:26 Got the following output from it: 2015-09-26 11:39:27 * Caching service dependencies ... [ ok ] 2015-09-26 11:39:30 * no services specified 2015-09-26 11:39:40 <_xs> yep, hava a loog at the bottom of the page here http://www.funtoo.org/Package:OpenRC 2015-09-26 11:39:44 <_xs> sorry, no anchors 2015-09-26 11:39:52 <_xs> "OpenRC is ignoring dependencies and not reacting to dependency changes" section 2015-09-26 11:40:52 <_xs> look* 2015-09-26 11:41:22 Trying, I'll let you know if it works... 2015-09-26 11:41:56 Nope, I still have the same problem. 2015-09-26 11:42:16 :( 2015-09-26 11:42:26 <_xs> ensure dbus is in default group, regenerate the cache and reboot 2015-09-26 11:43:26 <_xs> sorry isn't dbus but udev in fault 2015-09-26 11:43:43 <_xs> just checked now in the setup-xorg-base script 2015-09-26 11:43:45 That would make more sense to me 2015-09-26 11:44:14 Is udev specifically needed rather than mdev? 2015-09-26 11:44:26 <_xs> it runs rc-update -q add udev sysinit which doesn't work due to the cache problems 2015-09-26 11:44:44 <_xs> I think Xorg don't work with mdev 2015-09-26 11:45:34 :( 2015-09-26 11:46:06 <_xs> yup :/ 2015-09-26 11:47:31 Ah, installing udev made it work. 2015-09-26 11:47:46 I didn't need to do anything with the cache, just installing and starting it was enough. 2015-09-26 11:47:56 <_xs> cool 2015-09-26 11:47:56 Though maybe after a boot I will need to mess with the cache. 2015-09-26 11:48:08 <_xs> maybe 2015-09-26 11:48:22 <_xs> got that issue from the 3.1 or near 2015-09-26 11:48:56 I'll watch out for it. Thanks. 2015-09-26 11:49:12 <_xs> you're welcome 2015-09-26 11:49:45 <_xs> sorry for before, i'm tired helping my sister to clear off at the same time and she isn't really enjoyable.. 2015-09-26 11:49:52 <_xs> so must go, good afternoon 2015-09-26 11:50:52 you too 2015-09-26 11:51:02 good luck 2015-09-26 13:24:32 Hello good people 2015-09-26 13:25:01 i have been playing with alpine and i was able to make it work xhyve 2015-09-26 13:25:15 im having a small issue with the network 2015-09-26 13:25:34 when im installing my network works well 2015-09-26 13:26:19 but once installed im receiving this error 2015-09-26 13:26:25 ip: ioctl 0x8913 failed: No such device 2015-09-26 13:26:25 ip: can't find device 'eth0’ 2015-09-26 13:26:47 does any one have stumble up across something like this? 2015-09-26 14:46:27 argami: iv'e seen strange network device nmes. maybe see "ifconfig -a" 2015-09-26 14:48:27 blackpenguin: thanks for answer i alrady look on that but i just can see the lo 2015-09-26 14:48:40 the error is still the same as before 2015-09-26 14:49:01 what is more strange is that when i load alpine from cd 2015-09-26 14:49:11 and configure the network 2015-09-26 14:49:15 it works well 2015-09-26 14:49:41 but the problem comes after i install in hdd and start from the hdd 2015-09-26 14:49:58 not sure if the issue is alpine or xhyve 2015-09-26 14:53:33 got the name of the module needed in /etc/modules? 2015-09-26 18:30:53 Hello o/ 2015-09-26 18:32:29 I'm here couse I found logs of discussion about grsecurity thing ;-) 2015-09-26 18:33:23 I've been using alpine few times - in fact I rly like it, but I've got alot of servers with debian and centos with custom grsec kernel 2015-09-26 18:33:29 stable branch of course 2015-09-26 18:35:02 so grsec *thing* also impact on me.. 2015-09-26 18:37:17 so I'm curious - have you found some solution? Are you gonna drop grsecurity or move to testing branch? 2015-09-26 22:09:56 hi all! I’m trying to install a package which appears to have been moved from testing to main yesterday. it’s no longer in any of the testing mirrors, but doesn’t seem to have hit the main mirrors yet… any idea how long things usually take to propagate through the mirrors 2015-09-26 22:15:31 pr3hensile: are you on stable 2015-09-26 22:15:45 ammunta: yep 2015-09-26 22:15:56 you'll find it in edge 2015-09-26 22:19:54 ammunta: ah yes, gottit. thanks very much :) 2015-09-26 23:57:11 anyone use 3.2.3x64 with pyinstaller 2.1 successfully? ERROR: Can not find libc.musl-x86_64.so.1 in path ldd. So, won't work with alpine musl libc? 2015-09-27 00:30:49 wait what? 2015-09-27 00:31:29 what is pyinstaller? I can't seem to find a pyinstaller package in main/ or testing/ 2015-09-27 00:32:00 also: what exact command did you execute to see that error? 2015-09-27 00:33:48 installed via "pip install pyinstaller" A program that packages Python programs into stand-alone executables, under Windows, Linuxaller" it's a 2015-09-27 00:36:24 neat little app, but never tried in alpine before (just discovered alpine a few weeks ago) 2015-09-27 00:37:36 pyinstaller -y -F ./Meta-Man.Server/main_slave.py 2015-09-27 00:37:39 to build the app 2015-09-27 18:17:17 Hi, is this a roling distro ? 2015-09-27 18:17:30 <_ikke_> No 2015-09-27 18:17:31 I mean I can safely upgrade to future versions just with apk 2015-09-27 18:17:34 I see 2015-09-27 18:17:54 ok but I can still upgrade without update media etc 2015-09-27 18:17:54 <_ikke_> They have fixed releases, but you can update using just APK 2015-09-27 18:18:00 great 2015-09-27 18:18:39 Roland-: Alpine-linux got much more frequent releases than most other distros, in my opinion. So it is quite nice and "almost" rolling 2015-09-27 18:19:04 yes I am mostly interested just not to pull out a big upgrade like debian 7 to 8 and stff 2015-09-27 18:19:19 apk update weekly should keep me 2015-09-27 18:20:34 I am setting an almost embedded small box to do just firewalling 2015-09-27 18:20:36 shorewall 2015-09-27 18:20:46 since distro is lighweight, won 2015-09-27 18:22:36 <_ikke_> For new releases, you do have to update the repository too 2015-09-27 18:22:53 new I mean 2015-09-27 18:22:56 3.2 such 2015-09-27 18:23:05 <_ikke_> If you follow edge, you do get a rolling release 2015-09-27 18:23:06 major 2015-09-27 18:25:15 seems like parted nor cfdisk are included 2015-09-27 18:25:35 initial iso 2015-09-27 18:25:47 <_ikke_> No, default install is very lean 2015-09-27 18:33:27 nice to read if dealing with json, https://tools.ietf.org/html/rfc6901, http://jsonapi.org/format/ 2015-09-28 06:55:37 morning 2015-09-28 08:11:35 moin 2015-09-28 08:27:31 <_xs> moinmoin 2015-09-28 10:27:51 What are the chances I can get zfs working this week? 2015-09-28 11:59:51 hi there :) 2015-09-28 12:49:53 can anyone try inkscape in Alpine Linux 3.2? 2015-09-28 13:00:02 does it seg fault for everyone, or just me 2015-09-28 13:36:38 apk update && apk upgrade fixed inkscape :) 2015-09-28 13:42:49 hmm, it worked once, not is seg faults each time 2015-09-28 14:03:00 ScrumpyJack: as usual, do you happen to see anything interesting in dmesg? 2015-09-28 14:03:20 because in 95% of cases i find a grsec like message there but don't find it for days 2015-09-28 14:03:35 oh and on osx it ALLLLLWAAAAAYYSS segfaults for me this is why i stopped using it 2015-09-28 14:03:45 weird font issues were the cause in that case 2015-09-28 14:09:26 nothing interesting in /var/log/messages 2015-09-28 14:14:23 clandmeter: about lxd. i get http://tpaste.us/G5Ex 2015-09-28 14:15:00 ScrumpyJack: sounds fun 2015-09-28 14:25:08 does lxd work for you? 2015-09-28 14:26:03 i didnt really try 2015-09-28 14:26:10 i think i tried --help or something 2015-09-28 14:28:49 :) 2015-09-28 16:39:11 what's "community"? 2015-09-28 18:18:35 ammunta: software currently in main which is impossible to maintain long term support for 2015-09-28 20:27:29 Found this on reddit, it might interest you. https://www.reddit.com/r/linux/comments/3mqqtx/alpine_linux_why_no_one_is_using_it/ 2015-09-28 22:52:09 grsec and sandboxed apps by default 2015-09-28 22:52:13 sandboxed ? 2015-09-28 23:09:52 what package is libcurl? 2015-09-28 23:10:02 curl-dev? 2015-09-28 23:59:51 just apk add curl.......unless u need the development headers/source 2015-09-29 00:00:08 spoot, thanks. got it 2015-09-29 01:31:44 I believe most of those are fake 2015-09-29 03:05:19 haha san francisco 2.0 2015-09-29 03:19:51 Hi, need some advice. An `apk fix` call failed on one package due to outdated local cache, and now all apk commands result in: https://gist.github.com/kabaka/6882af097811acc5d410 2015-09-29 03:21:22 I added -f (and -s, of course) to several commands I thought might get me out of this, but the output is that all packages would be purged. 2015-09-29 09:36:32 heh, colleague gave me the go-ahead to port our shit from openwrt to alpine 2015-09-29 09:37:43 Diftraku, cool! 2015-09-29 09:40:53 fcolista: we've slowly been getting fed up with uclibc+busybox and lacking any sane environment 2015-09-29 09:41:44 Kabaka: there's a apk static 2015-09-29 09:41:48 err apk-static 2015-09-29 09:42:10 it's also available in a tarball iirc, so you can fetch it and fix things with it 2015-09-29 10:13:02 Diftraku: note that openwrt trunk now is musl based 2015-09-29 10:30:57 uggedal: that still doesn't fix the insanity that is the environment :3 2015-09-29 10:31:07 it's "special" at best 2015-09-29 11:28:23 it really seams like git is releasing a new version every few days now 2015-09-29 11:31:08 leo-unglaub: it's so it can stay "hip" 2015-09-29 11:31:19 hehe, yeah 2015-09-29 11:31:26 version number race FTW! 2015-09-29 11:42:13 just like windows or firefox 2015-09-29 12:47:28 Does alpine run on those cheap routers with the (I think) Broadcom chipsets? 2015-09-29 12:48:25 afaik only arm or x86 so far 2015-09-29 15:36:08 is there a document anywhere on bootstrapping alpine from source from scratch, e.g. for a new target arch? 2015-09-29 15:39:56 dalias, not really. the scripts i've used are online at: http://dev.alpinelinux.org/~tteras/bootstrap/ 2015-09-29 15:40:02 though, it's been a while since last bootstrap 2015-09-29 15:40:12 as you see from script timestamps :) 2015-09-29 15:43:51 i'm trying to convince landley we could do alpine for sh2 :-) 2015-09-29 15:49:53 looks pretty simple 2015-09-29 15:50:05 so once you get that far, everything else is compiled native, right? 2015-09-29 15:52:03 yes 2015-09-29 15:52:21 it crossbuilds build-base which enough to rebuild everything natively 2015-09-29 15:52:38 give or take. we occationally get cyclic build dependencies, but try to fix those regularly 2015-09-29 15:52:51 <_ikke_> Is there a difference between cross-build and native? 2015-09-29 15:53:38 yes 2015-09-29 20:10:55 does anyone know how i can get data rates with ifconfig 2015-09-29 20:10:57 as a user 2015-09-29 20:11:11 rx/tx shows up as 2015-09-29 20:11:14 RX bytes:0 (0.0 B) TX bytes:0 (0.0 B)\ 2015-09-29 20:12:36 or for that matter 2015-09-29 20:12:39 view cat /sys/class/net/eth0/statistics/rx_packets 2015-09-29 20:12:41 as a user 2015-09-29 20:13:51 main reason is im using conky 2015-09-29 20:13:55 but i didn't want to run it as root 2015-09-29 20:16:39 Hi folks, is it possible to rebuild pkg for another architecture? eg my APKBUILD is x86_64 and i need it for arm..?! sorry, im new to alpine ;-) 2015-09-29 20:16:59 sillysausage, I believe that's grsecurity related 2015-09-29 20:17:12 ah you mightbe right icicled 2015-09-29 20:17:15 sillysausage: reading that doesn't need root privileges, but ifconfig only shows accumulated data, you would need to compare that to a former value and time to calculate a rate 2015-09-29 20:17:17 <_ikke_> sattler: Do you know how to build packages? 2015-09-29 20:17:21 abuild -c -r -P /target 2015-09-29 20:17:23 Package not available for the target architecture (armhf). Aborting. 2015-09-29 20:17:35 i was able to fix it by running conky as root 2015-09-29 20:17:37 <_ikke_> sattler: You can try to just add armhf to the list 2015-09-29 20:17:38 but i dont want to do that 2015-09-29 20:17:49 also it shows 0 packets in /sys/class/net/eth0/statistics/rx_packets 2015-09-29 20:17:53 <_ikke_> sattler: that's what I usually try 2015-09-29 20:17:59 unless i actually run it with sudo (just catting) 2015-09-29 20:18:05 so its obviously i cannot do it as aregular user 2015-09-29 20:18:14 is there any way i can get around that 2015-09-29 20:18:40 sillysausage: you shouldn't need to do that. adding sbin to your path or calling ifconfig as /sbin/ifconfig should be enough 2015-09-29 20:18:42 some sysctl? 2015-09-29 20:18:51 glibc for armhf is not available...?! looked around and had no luck :-( 2015-09-29 20:19:08 blackpenguin: nah itsnot i think it is what icicled said 2015-09-29 20:19:14 because cat /sys/class/net/eth0/statistics/rx_packets 2015-09-29 20:19:17 says permission denied 2015-09-29 20:19:28 I had a similar issue when trying to view battery stats in alpine 2015-09-29 20:19:35 yeah 2015-09-29 20:19:46 I never did solve it though =\ 2015-09-29 20:19:55 must be grsec rules then (I dont use the grsec kernel) 2015-09-29 20:20:00 <_ikke_> sillysausage: what version of alpine? Because since 3.0 alpine uses musl 2015-09-29 20:20:10 <_ikke_> sattler: * 2015-09-29 20:20:21 the latest err 3.2.3 2015-09-29 20:20:41 oh :P 2015-09-29 20:20:54 sillysausage, you'll want to disable "Sysfs/debugfs restriction" 2015-09-29 20:20:56 <_ikke_> autocomplete fail :P 2015-09-29 20:21:00 icicled: ah 2015-09-29 20:21:14 i shouldn't need to recompile the kernel? 2015-09-29 20:21:21 I hope not :P 2015-09-29 20:22:00 im hoping there will be a nice sysctl 2015-09-29 20:22:17 https://en.wikibooks.org/wiki/Grsecurity/Appendix/Grsecurity_and_PaX_Configuration_Options#Sysfs.2Fdebugfs_restriction 2015-09-29 20:23:19 on my desktop i took the pkg.tar.xz from the arch-guys but they do not have a pkg for armhf :-( 2015-09-29 20:23:33 its probably better to run conky as root 2015-09-29 20:23:44 than it is to disable that grsec kernel setting 2015-09-29 20:25:07 sillysausage, unfortunately I think you'll have to re-compile the kernel with GRKERNSEC_SYSFS_RESTRICT disabled 2015-09-29 20:25:15 yeah ill just run it as root 2015-09-29 20:25:24 imo that's safer 2015-09-29 20:25:34 ehhh 2015-09-29 20:25:34 disabling that means everything can then read sysfs 2015-09-29 20:25:35 Oo 2015-09-29 20:25:50 where as if i run conky as root 2015-09-29 20:25:55 is it possible to change the group permissions on that sysfs path? 2015-09-29 20:25:56 then only it can 2015-09-29 20:26:00 or that 2015-09-29 20:26:11 <_ikke_> icicled: Doesn't help, it's blocked by grsec 2015-09-29 20:26:42 and also owned by root:root anyway 2015-09-29 20:27:18 add a line to rc.local that changes the permission on that path 2015-09-29 20:27:38 well that would work 2015-09-29 20:27:47 <_ikke_> would it? 2015-09-29 20:27:52 idon't know 2015-09-29 20:27:54 idk try it! 2015-09-29 20:28:06 well i can try to just sudo chown it 2015-09-29 20:28:07 right now 2015-09-29 20:28:08 <_ikke_> I think it checks uid 2015-09-29 20:28:11 yep go for it 2015-09-29 20:28:39 yeah doesn't change it 2015-09-29 20:29:26 oh wait oops 2015-09-29 20:29:28 i made a mistake 2015-09-29 20:30:02 nahit doesn't 2015-09-29 20:30:08 $ sudo chown -R user:user /sys/class/net/eth0/statistics/rx_bytes 2015-09-29 20:30:12 $ cat /sys/class/net/eth0/statistics/rx_bytes 2015-09-29 20:30:16 cat: /sys/class/net/eth0/statistics/rx_bytes: Permission denied 2015-09-29 20:30:26 -r--r--r-- 1 user user 4096 Sep 30 05:56 rx_bytes 2015-09-29 20:30:33 and it shows its owned by user:user 2015-09-29 20:30:37 so obviously still blocked 2015-09-29 20:30:41 o_O 2015-09-29 20:30:48 <_ikke_> This is not blocked by file permissions 2015-09-29 20:30:53 sillysausage: any messages in dmesg? 2015-09-29 20:30:54 yeah that wouldn't work 2015-09-29 20:30:58 <_ikke_> It just hard checks if your UID is 0 2015-09-29 20:31:05 nope 2015-09-29 20:31:08 true 2015-09-29 20:31:15 and i can't make my uid 0 2015-09-29 20:31:20 so i dont see what i could add to rc.local 2015-09-29 20:31:22 sillysausage, could you create a wrapper script & have conky point to that? 2015-09-29 20:31:30 well that's an idea 2015-09-29 20:31:32 to read that one val from /sys 2015-09-29 20:31:44 and could just echo it 2015-09-29 20:31:46 the script could be setuid 2015-09-29 20:31:54 that's a good idea 2015-09-29 20:32:09 actually, you can't setuid scripts 2015-09-29 20:32:19 <_ikke_> indeed 2015-09-29 20:32:27 off to c! 2015-09-29 20:32:29 <_ikke_> You can, but it won't apply 2015-09-29 20:32:34 skarnet: you can setuid /bin/sh :P 2015-09-29 20:32:42 lol 2015-09-29 20:32:44 XD 2015-09-29 20:32:45 <_ikke_> uggedal: Sounds like a great plan 2015-09-29 20:33:21 I would consider to run a script that reads from /sys and puts it public readable somewhere else as root (instead of running conku as root) 2015-09-29 20:33:44 in university, I once exploited a NFS root bug and made a setuid copy of /bin/sh to my $HOME/private/bin directory. I called it "please". 2015-09-29 20:33:57 please cat /etc/shadow 2015-09-29 20:34:06 was fun 2015-09-29 20:34:06 <_ikke_> lol 2015-09-29 20:34:16 much easier to just create a tiny c program that reads from that file to stdout, compile it, and setuid it 2015-09-29 20:34:33 <_ikke_> yeah 2015-09-29 20:34:53 sure, but I was just starting to learn Unix at the time. Now I'd do that, obviously. 2015-09-29 20:35:21 was talking about the net statistics gathering for conky :) 2015-09-29 20:35:32 ah, my bad XD 2015-09-29 20:36:09 but ifconfig *should* print stuff even without root permissions 2015-09-29 20:36:27 <_ikke_> you can setuid ifconfig :P 2015-09-29 20:37:09 there's something wrong in the path to the stats, sysfs should be world-readable 2015-09-29 20:37:46 <_ikke_> skarnet: grsec blocks it 2015-09-29 20:38:06 ACTION facepalms 2015-09-29 20:38:48 "grsecurity, the hardening framework that makes you write setuid programs to work around it" 2015-09-29 20:38:56 remember grsec is the paranoid version of linux ;) 2015-09-29 20:40:32 <_ikke_> skarnet: Better then leaving everything wide-open, no? 2015-09-29 20:40:41 re: GRKERNSEC_SYSFS_RESTRICT, it is "mostly accessible only by root" 2015-09-29 20:41:10 the following are available to use are regular users: /sys/fs/selinux, /sys/fs/fuse, /sys/devices/system/cpu 2015-09-29 20:41:11 depends on the context really. 2015-09-29 20:45:12 skarnet: is s6-linux-init getting s6-rc support at some point? 2015-09-29 20:45:58 uggedal: doesn't need to: the s6-rc calls are supposed to happen in your stage 2 script. 2015-09-29 20:46:16 skarnet: ah, ok. thanks 2015-09-29 20:46:34 your stage 2 script can basically be "s6-rc-init /run/service && s6-rc change my-bundle" 2015-09-29 20:47:25 we're thinking of bringing an alternative init/supervision suite to alpine 2015-09-29 20:47:32 are you? 2015-09-29 20:47:37 yes 2015-09-29 20:47:45 well I wrote to ncopa today about it XD 2015-09-29 20:47:50 well, been thinking for years 2015-09-29 20:48:11 I'm willing to give all the necessary support if you want to bring s6 in 2015-09-29 20:48:25 thanks 2015-09-29 20:48:45 I was initially thinking of runit (that was 1-2 years ago) 2015-09-29 20:49:06 I guess. Void Linux uses runit, too. 2015-09-29 20:49:11 yep 2015-09-29 20:49:31 s6-rc seems really interesting 2015-09-29 20:50:41 honestly, s6 already had interesting features before s6-rc, but they went mostly under the radar 2015-09-29 20:51:32 yeah, my impression was that s6 is "better", but honestly I was a bit put off by execline 2015-09-29 20:52:01 there's no reason to, you can mostly ignore execline 2015-09-29 20:52:10 and write your run scripts in sh or whatever 2015-09-29 20:52:37 yeah, runscripts should mostly be oneliners anyways so don't matter what the shebang is 2015-09-29 20:53:12 I use execline everywhere because it's much easier to autogenerate than sh 2015-09-29 20:53:37 but for hand-written scripts, any language goes 2015-09-29 20:53:44 one sticky point for alpine would be how to handle /etc/conf.d 2015-09-29 20:54:06 I think ncopa wanted support for it (maybe sharing it with openrc) 2015-09-29 20:54:36 which would work ok for simple things I guess 2015-09-29 20:55:36 that would require some initial work in the scripts, but nothing dealbreaking 2015-09-29 20:56:21 just requires using sh for sourcing stuff in conf.d 2015-09-29 20:57:27 yeah, or create a wrapper binary to use for #! in run scripts which provides settings from conf.d in the env 2015-09-29 20:58:01 that's a possibility, but I was thinking more about oneshots 2015-09-29 20:59:36 either you store the whole script in a global place such as /etc/init.d and you write "/etc/init.d/foo start" in the "up" script 2015-09-29 21:00:30 would need to disable forking daemons, no? 2015-09-29 21:00:33 or you make the "up" (and "down") script local, but it requires a bit of manipulation: /bin/sh -c ". /etc/conf.d/foo; script" 2015-09-29 21:00:41 or does s6 do some cgroups woodo? 2015-09-29 21:01:09 no cgroups voodoo, but there are 2 ways to make it work: 2015-09-29 21:01:50 1. fast, easy way, doable as a first step: use your existing rc scripts as "oneshot" services, don't supervise the daemons. No changes to the scripts to speak of 2015-09-29 21:02:20 at least from my perspective I would like to do it properly with custom run scripts for every daemon 2015-09-29 21:02:32 2. better, but requires more work, can be done incrementally later: convert daemons to "longrun" services, using supervision. There you need to un-background them. 2015-09-29 21:03:09 I agree 2. is the proper way, but as a proof-of-concept you can start with 1. 2015-09-29 21:03:50 well, I run 3 daemons on this system, so should not be hard to write run scripts for them :) 2015-09-29 21:04:10 heh :) 2015-09-29 21:04:56 but I like that s6-rc provides oneshots. in Void we had a few run scripts that did their work and called pause 2015-09-29 21:05:09 some guy has whipped up a JS VM with s6-rc for people who want to try it out, let me find the URL 2015-09-29 21:05:23 like alsactl/iptables 2015-09-29 21:05:46 http://linsam.homelinux.com/extra/s6-altsim/jor1k/demos/s6-rc.html 2015-09-29 21:05:59 ah, you worked on Void too? 2015-09-29 21:06:13 yes, use it on desktop 2015-09-29 21:06:58 and yes, it was a recurrent point on the supervision mailing-list, people who were trying to use supervision as a service manager, with scripts that looped or paused 2015-09-29 21:07:13 that was ugly and kludgy 2015-09-29 21:08:40 I have to call it a night. idling in our channel will win major points when we start looking at alternative init after the pending 3.3 release is branched :) 2015-09-29 21:09:05 ETA? 2015-09-29 21:09:11 1 oct 2015-09-29 21:09:12 so that I know when to lurk around 2015-09-29 21:09:24 ok 2015-09-29 21:09:28 well I'll be there :) 2015-09-29 21:10:19 I have started some work locally, to make openrc optional. will be a bit of work like that first. but a prototype with a custom init= kernel cmdline would also be nice to get going 2015-09-29 21:10:57 trivial to do with s6-linux-init! :P 2015-09-29 21:11:17 less trivial to integrate with the way Alpine does stuff, I suppose 2015-09-29 21:11:45 but that's just reusing frameworks and modifying scripts. 2015-09-29 21:12:25 if I'm not on the channel when you guys discuss, just send me a mail, or a tweet. I'll get it and log in. 2015-09-29 21:29:48 hmm im thinking cpu percentage is probably also restricted 2015-09-29 21:29:50 for each process 2015-09-29 21:30:49 yeah it is. 2015-09-29 21:31:25 though im able to see memory and pid 2015-09-29 21:38:21 anyone know right off how to make openrc (or whatever is handling it) ignore the power button? 2015-09-30 00:28:19 alright, if I wanted to remaster the ISO and include additionals packages into the boot environment, where's the best place to start? 2015-09-30 00:32:33 go keeps installing at version 1.4 2015-09-30 00:32:45 even though 1.4.1-r2 should be in the repo 2015-09-30 00:32:59 I tried --repository http://nl.alpinelinux.org/alpine/edge/main 2015-09-30 00:36:26 repo is 1.5.1-r2 2015-09-30 00:37:37 nevermind got it :) 2015-09-30 00:41:02 atomi: What did you have to do? 2015-09-30 08:15:07 morning 2015-09-30 08:21:01 <^7heo> tnt bom bom FTW! 2015-09-30 08:21:05 <^7heo> yaaay! 2015-09-30 08:21:08 dalias: to make openrc ignore power button you can either "rc-service acpid stop" or comment out poweroff in /etc/acpi/PWRF/00000080 2015-09-30 08:21:20 <^7heo> #SuchIsolation 2015-09-30 08:21:22 <^7heo> #WOW 2015-09-30 08:21:50 <^7heo> moin ScrumpyJack, ncopa 2015-09-30 08:22:02 morning ^7heo 2015-09-30 08:23:08 <^7heo> ncopa: wouldn't acpid stop affect more than just the power button? 2015-09-30 08:23:43 i think the only thing it currently handles is power button 2015-09-30 08:23:53 find /etc/acpi 2015-09-30 08:23:55 <^7heo> ok 2015-09-30 08:24:21 <^7heo> yep here, only power button. 2015-09-30 12:38:42 http://bedrocklinux.org/faq.html sounds a bit like how i use alpinelnux 2015-09-30 12:38:55 i have debian and archlinux kvms 2015-09-30 12:41:41 Ooo, very interesting 2015-09-30 12:52:39 great, another level of abstraction 2015-09-30 12:52:59 and indirection 2015-09-30 13:27:44 https://packages.debian.org/jessie/printer-driver-gutenprint 2015-09-30 13:27:56 how would it best be to package "gutenprint" for alpinelinux? 2015-09-30 13:28:12 there seems to be a number of drivers that are needed 2015-09-30 13:28:20 err packages 2015-09-30 13:28:25 https://packages.debian.org/search?searchon=names&keywords=gutenprint 2015-09-30 13:28:46 only reason I ask is because currently my driver is unavailable 2015-09-30 13:29:07 on debian i install printer-driver-gutenprint (which pulls in libgutenprint2) 2015-09-30 13:30:45 https://dpaste.de/Uz6c 2015-09-30 13:31:28 probably someone with advanced APKBUILD knowledge should do this ive not packaged anything like this before 2015-09-30 13:31:58 https://bugs.alpinelinux.org/issues/4656 i filed a bug here a few days ago, but I'm not sure really how to do this 2015-09-30 13:32:13 would we split it up the same way as debian? 2015-09-30 13:33:53 because it comes as a single distribution package http://downloads.sourceforge.net/project/gimp-print/gutenprint-5.2/5.2.10/gutenprint-5.2.10.tar.bz2 2015-09-30 13:35:44 https://wiki.gentoo.org/wiki/Printing#Using_Special_Printer_Drivers 2015-09-30 13:37:18 and it seems my printer isn't supported by https://wiki.gentoo.org/wiki/HPLIP 2015-09-30 13:38:46 https://packages.gentoo.org/package/net-print/gutenprint seems gentoo have one package 2015-09-30 13:43:00 hmm yeah hplip seems to be mostly for multi function printers 2015-09-30 13:44:41 https://bugs.alpinelinux.org/issues/4656 anyway i filed a bug for it i know that's not really in the spirit of things 2015-09-30 14:06:50 brb 2015-09-30 14:29:13 ncopa, ^7heo, eh? it looks like it also does suspend-on-lid 2015-09-30 14:55:03 dalias: oh, yes, thats true 2015-09-30 14:55:39 i'd recommend comment out the line in /etc/acpi/PWRF/00000080 2015-09-30 14:56:05 you can also chmod -x it but then i think apk fix or next apk upgrade will restore the permissions 2015-09-30 16:09:27 ncopa: did you see my messages about gutenprint? 2015-09-30 17:12:32 i hate the patch command 2015-09-30 17:12:43 applying patches is so fucking hard to get right with that syntax 2015-09-30 17:12:47 i have to google that every time ... 2015-09-30 17:13:05 <_ikke_> I use it almost every time 2015-09-30 17:13:47 _ikke_: then you can quickly paste me a snippet to apply all patch files in the current directory? 2015-09-30 17:14:26 <_ikke_> leo-unglaub: sorry, I was thinking I was in a different channel :P 2015-09-30 17:14:33 hahaha 2015-09-30 17:14:46 <_ikke_> I always copy that from another apkbuild 2015-09-30 17:14:52 *g* 2015-09-30 17:15:13 see, the syntax is not that intuitive if you have to copy it every time *g* 2015-09-30 17:15:24 <_ikke_> It's not syntax 2015-09-30 17:15:27 <_ikke_> it's an idiom though 2015-09-30 17:16:56 <_ikke_> It's just that I'm too lazy to type it everytime 2015-09-30 17:18:50 <_ikke_> for i in source; do case i in; *.patch) apply -p1 $i;; esac; done 2015-09-30 17:18:53 <_ikke_> something like that 2015-09-30 17:23:30 ncopa: i didn't see if you replied to me 2015-09-30 17:23:48 switched computer, and forgot to check irc before i quit it 2015-09-30 17:26:47 <_ikke_> sillysausage: He did not reply after your last message 2015-09-30 17:26:52 ah kk 2015-09-30 17:27:11 i guess nobody uses printers these days ;) 2015-09-30 17:41:32 leo-unglaub: usually you just want patch -p1 < 2015-09-30 17:42:11 but you can also just use git-apply ... 2015-09-30 17:42:31 " 2015-09-30 17:42:31 Reads the supplied diff output (i.e. "a patch") and applies it to files. With the --index option the patch is also applied to the index, and with the --cached option the patch is only applied to the index. Without these options, the command applies the patch only to files, and does not require them to be in a Git repository. 2015-09-30 17:42:36 " 2015-09-30 18:01:47 dalias, btw. the minimal notes for alpine bootstrap are at: http://wiki.alpinelinux.org/wiki/User:Fabled 2015-09-30 18:15:23 fabled, thanks 2015-09-30 18:16:26 fabled, i see fakeroot is in the base packages 2015-09-30 18:16:47 are you aware alpine's fakeroot is broken? it fails to intercept mknod and *stat 2015-09-30 18:17:02 i noticed there's issues with it 2015-09-30 18:17:13 and the base package set may have changed since last bootstrap 2015-09-30 18:18:43 i have a patch but it breaks glibc 2015-09-30 18:18:51 it needs to be adapted to go upstream but alpine could use it 2015-09-30 18:21:21 please send to mailing list or open bug in our tracker. i'm about to go sleep soon. 2015-09-30 18:23:26 ok 2015-09-30 18:31:09 submitted 2015-09-30 18:31:26 http://bugs.alpinelinux.org/issues/4701 2015-09-30 18:38:38 I don't understand, I am part of the video group but I still get "libGL error: failed to open drm device: Permission denied" after logging in as a normal user and using startx 2015-09-30 18:40:08 has anyone seen that before? 2015-09-30 18:41:15 fabled, it will be interesting to see if these base packages support nommu... 2015-09-30 18:43:41 Karlton: check if you see an error in dmesg 2015-09-30 18:47:14 with grsec we should probably add "something not working? check dmesg" in /topic ... 2015-09-30 18:47:53 darkfader: anyone think I see that is unusual is "[ 1935.437985] udevd[11982]: specified group 'input' unknown" 2015-09-30 18:49:28 err *the only thing 2015-09-30 18:49:42 Karlton: what are the permissions on /dev/dri/card0 or whatever its called these days 2015-09-30 18:50:06 uggedal: thanks, i couldn't remember 'dri' so i shut up in desparation :) 2015-09-30 18:51:03 crw-rw---- 1 root root 226, 0 Sep 30 12:22 /dev/dri/card0 2015-09-30 18:51:43 wait, do I need a patched kernel? 2015-09-30 18:53:13 Karlton: no, but your user really doesn't have permssions on the card, see it? :) 2015-09-30 18:53:53 i think alpine doesn't set those up by default :> 2015-09-30 18:57:31 okay so I added /dev/dri/card0 to video group so it works for now :) 2015-09-30 21:05:33 ahoi 2015-09-30 21:05:34 ERROR: unsatisfiable constraints: 2015-09-30 21:05:34 chromium (missing): 2015-09-30 21:05:34 required by: world[chromium] 2015-09-30 21:05:56 how can i install chromium? 2015-09-30 21:06:18 edge and edge/testing repo added 2015-09-30 21:08:41 make a APKBUILD for it 2015-09-30 21:09:39 ? 2015-09-30 21:09:45 https://bugs.alpinelinux.org/issues/285 2015-09-30 21:10:03 clandmeter says "Its available in edge/main and will probably be included in next minor alpine version. 2015-09-30 21:12:07 it is indeed 2015-09-30 21:12:25 http://git.alpinelinux.org/cgit/aports/tree/main/chromium?id=61dd7a457049d16ed051eb2f693dca4950af469b 2015-09-30 21:12:34 kocka: add community repo 2015-09-30 21:22:10 hmm 2015-09-30 21:22:23 couldn't find any info on community repo 2015-09-30 21:25:25 has anyone here used Alpine Linux to build LFS with musl libc? 2015-09-30 21:26:13 I already failed at GCC pass 1 with a stupid error 2015-09-30 21:26:21 "../../gcc-5.2.0/gcc/config/linux.h:95:30: error: expected ',' or ';' before 'MUSL_DYNAMIC_LINKER32'" 2015-09-30 21:27:36 even though there is a , before 'MUSL_DYNAMIC_LINKER32' ; I checked the source myslef 2015-09-30 23:43:46 s6 is in main and runit's still in testing, huh