2015-06-01 01:43:24 hello, has anyone successfully compiled ruby 2.2.0 on alpine docker? 2015-06-01 01:43:34 I'm getting "/usr/lib/gcc/x86_64-alpine-linux-musl/4.9.2/../../../../x86_64-alpine-linux-musl/bin/ld: ./libffi-3.2.1/.libs/libffi.a(raw_api.o): relocation R_X86_64_PC32 against symbol `ffi_ptrarray_to_raw' can not be used when making a shared object; recompile with -fPIC" 2015-06-01 01:43:45 "/usr/lib/gcc/x86_64-alpine-linux-musl/4.9.2/../../../../x86_64-alpine-linux-musl/bin/ld: final link failed: Bad value" 2015-06-01 02:26:34 why are you building it, it's a package 2015-06-01 02:26:49 so yes the automatic build systems have done it ;) 2015-06-01 02:32:36 I'm building images with multiple versions of ruby, for this one I'm keeping it using ruby-install for the consistency of my other images 2015-06-01 02:32:45 and so when the package changes, this image still installs 2.2.0 2015-06-01 02:46:12 delfick: check out the aport for ruby and see if it has any special configuration in the build() function 2015-06-01 02:46:28 APKBUILDs are straightforward recipes since they are literally shell scripts 2015-06-01 02:49:09 ahills: ok, sure, where can I find that? 2015-06-01 02:50:05 delfick: http://git.alpinelinux.org/cgit/aports/tree/main/ruby/APKBUILD 2015-06-01 02:50:44 thankyou 2015-06-01 02:59:11 no problem 2015-06-01 03:16:57 what's the linux-virtgrsec package for? 2015-06-01 03:17:03 xen hosts/guests? 2015-06-01 03:26:02 man, I'm having some wild issues with my xen dom0 after updating to alpine 3.2 (xen 4.5) 2015-06-01 03:26:14 page faults all over the place, BUG lines in dmesg followed by stack traces... 2015-06-01 03:26:26 I'm glad I migrated my important stuff to containers 2015-06-01 03:46:11 xen is really kind of going downhill 2015-06-01 03:46:42 I think it's miserable but I'm already locked in 2015-06-01 03:46:48 unless there's an easy way to switch to kvm 2015-06-01 03:49:34 ahills: are they PV or HVM guests. 2015-06-01 03:49:39 ugh, kvm uses xml as configuration? 2015-06-01 03:49:46 kaniini: all PV 2015-06-01 03:50:00 I migrated from 3.1 to 3.2 2015-06-01 03:50:14 hmm. 2015-06-01 03:50:19 I'm slowly trying each one 2015-06-01 03:50:31 now, I've also recently added one--a debian jessie image 2015-06-01 03:50:44 what errors are you getting / have you tried using linux-vanilla instead of linux-grsec ? 2015-06-01 03:50:45 I'm going through and starting the guests one by one 2015-06-01 03:50:51 ohhh. 2015-06-01 03:50:56 I'm trying to capture the errors but it's very difficult 2015-06-01 03:50:59 you're probably seeing too much pressure on xenstore 2015-06-01 03:51:02 the *host* is the one freezing 2015-06-01 03:51:12 have you dedicated a CPU core to dom0? 2015-06-01 03:51:23 no, but I only have 8 CPUs 2015-06-01 03:51:36 you really should just eat the penalty 2015-06-01 03:51:41 it will make things a lot more stable 2015-06-01 03:51:47 haha, if it means I can use it at all 2015-06-01 03:52:08 but yes, libvirt is XML 2015-06-01 03:52:28 are there config tools for it, or are they ignoring the #1 bullet point of xml, which is that "it's not meant to be human-readable" 2015-06-01 03:52:55 i dont know. i moved that type of workload to ESXi 2015-06-01 03:53:22 which, admittedly, i am not too happy with 2015-06-01 03:54:14 I'm glad to not be using ESXi anymore 2015-06-01 03:56:51 God, I wish I could somehow capture these kernel panics 2015-06-01 03:57:20 is the only way to reserve a CPU for dom0 to also restrict it? 2015-06-01 03:57:41 yes, dom0_cpu_pin 2015-06-01 03:58:01 or I could just set the cpu affinity of every guest ;) 2015-06-01 04:17:33 well the problem is you're probably experiencing a deadlock where the soft-interrupt crosses the same core as dom0 2015-06-01 04:22:06 that seems like an avoidable deadlock to me, but I pinned the vcpu anyways and it seems to be stable now 2015-06-01 04:23:07 nah dom0 is really fragile 2015-06-01 04:23:10 if it lags 2015-06-01 04:23:13 everything goes to shit 2015-06-01 04:31:55 I'm worried that's still going to happen even with the pinning... at least the debian guests cause a lot of activity in their corresponding qemus 2015-06-01 04:32:11 to run a dozen guests requires nearly 4GB RAM in the dom0... 2015-06-01 04:34:15 that seems odd. i was running dom0 with 1GB RAM and 50-60 guests per node when i ran a hosting company. 2015-06-01 04:34:32 yeah, I think there is a problem with my setup 2015-06-01 04:35:08 either that, or the memory issues I was seeing were just subsymptoms of the CPU availability problem 2015-06-01 04:37:35 ahills: are you booting with dom0_mem=1G? 2015-06-01 04:37:48 dom0_mem=4096M 2015-06-01 04:37:52 ahills: if you use ballooning it will result in heap fragmentation and large overhead from that 2015-06-01 04:37:59 ballooning is disabled 2015-06-01 04:38:04 or should be... 2015-06-01 04:38:05 very odd then. 2015-06-01 04:38:12 try, 2015-06-01 04:38:16 I should check, the config format changes all the time 2015-06-01 04:38:17 dom0_mem=4096M,max:4096M 2015-06-01 04:38:25 isn't that the old syntax? 2015-06-01 04:38:37 no idea, but you want to set the max too 2015-06-01 04:38:47 otherwise it will assume it can balloon up to maximum 2015-06-01 04:38:59 and create pte's for it 2015-06-01 06:44:32 ahills: fwiw i updated one of my hosts to alpine-xen 3.2 and haven't had any of the problems you descibe 2015-06-01 06:51:44 morning folks 2015-06-01 07:13:47 ay coredumb 2015-06-01 07:14:26 <_ikke_> morning 2015-06-01 11:31:36 hello 2015-06-01 11:32:23 do alpine runs on the raspberry pi 2 ? 2015-06-01 11:34:13 Chipster1, yes 2015-06-01 11:34:25 nice 2015-06-01 11:34:48 I have it running, 2015-06-01 11:35:23 with which .tar.gz ? 2015-06-01 11:35:36 RPi or generic arm ? 2015-06-01 11:35:51 rpi 2015-06-01 11:35:56 excellent 2015-06-01 11:35:57 but I have a problem, only recognizes 768MB of RAM 2015-06-01 11:36:02 ok 2015-06-01 11:36:46 the rest is it used to video card ? 2015-06-01 11:37:05 s/to/by/ 2015-06-01 11:37:07 I guess so 2015-06-01 11:37:27 ok 2015-06-01 11:43:52 thx for your help 2015-06-01 11:49:43 <_ikke_> You should be able to config that in //boot/config.txt 2015-06-01 12:20:36 i know but there is some bugs 2015-06-01 12:55:17 Chipster1: some of it is used by the root filesystem / 2015-06-01 12:55:40 ScrumpyJack: yeah, I also had similar problems a while back, but thought I solved them by increasing dom0 memory allocation 2015-06-01 12:55:59 everything *seems* to be stable now 2015-06-01 12:57:12 glad to hear it 2015-06-01 13:04:59 does someone here use rrdtool on alpinelinux? 2015-06-01 13:05:11 crow: I use smokeping, which I think uses rrdtool as backend 2015-06-01 13:05:25 just a quick and dirty host heartbeat graph set 2015-06-01 13:05:35 because I have too many computers :( 2015-06-01 13:05:52 i need something to watch the RAM consumation, duno with what to use 2015-06-01 13:10:01 smokeping is a little hard to configure, but you can basically have it execute any command, so you could ssh free and process the output of that 2015-06-01 13:12:15 ahills ill check it, i dont want something hard, just to have basic info 2015-06-01 13:13:51 I think it's in aports 2015-06-01 13:13:55 I know rrdtool is 2015-06-01 13:17:33 wwwwwwww 2015-06-01 13:21:53 i use collectd for that on AL 2015-06-01 13:22:29 with collectd-network 2015-06-01 13:25:05 how have I never heard of this... this looks better than what I'm using 2015-06-01 13:26:48 the collectd-server runs in a LXC 2015-06-01 13:27:16 and collectd (client) in every machine be physical, virtual or container 2015-06-01 13:28:08 with cgp on the server it's really an esay view of the ressources usage for all tour machines 2015-06-01 13:34:02 <_ikke_> Do you also have something to collect / monitor logs? 2015-06-01 13:34:24 ACTION crosses fingers for a collectd debian package 2015-06-01 13:46:44 ahills: time to move your debian box to alpine-linux ;) 2015-06-01 13:47:28 yeah, I really want to 2015-06-01 13:47:32 it's a huge mess 2015-06-01 13:47:41 Jean-Scotch: do you collectd inside alpine linux LXCs? 2015-06-01 13:48:21 i mean running on alpine-linux host? 2015-06-01 13:48:24 for some if I need specific metric. otherwirhe the gost is good enough 2015-06-01 13:48:33 s/gost/host/ 2015-06-01 13:49:45 the server is an AL LXC on a AL host. The baremetal run the client as do the KVM and some LXC. 2015-06-01 13:50:49 I use collect on my debian boxes also 2015-06-01 13:51:13 did you build it? 2015-06-01 13:51:17 no 2015-06-01 13:51:32 from repo 2015-06-01 13:55:39 alpine-linux host only just got CONFIG_CGROUP stuff in kernel so not sure how accurate those stats would have been prior to 3.2 2015-06-01 14:03:06 Jean-Scotch do you have LXC container for the colectd-server maybe? 2015-06-01 14:05:59 crow: it cannot just being exported... because some keys are laying around 2015-06-01 14:07:24 but it was straight-forward to deploy. collectd collectd-network collectd-rddtool lightppd cgp 2015-06-01 14:09:17 shouldn't apk prefer a tagged repo for dependencies of a tagged package? 2015-06-01 14:09:34 This is somewhat ridiculous IMHO: 2015-06-01 14:09:45 Shinjuku:~# apk add samba-server@custom samba-common-tools@custom samba-libs@custom libsmbclient@custom libwbclient@custom samba-winbind-clients@custom samba-client@custom samba-winbind@custom samba-common-libs@custom samba-client-libs@custom samba-server-libs@custom 2015-06-01 14:09:56 crow: http://wiki.alpinelinux.org/wiki/Collectd 2015-06-01 14:15:05 Jean-Scotch thank you ill try it, i still dont have any lxc so thats a start point :) 2015-06-01 14:53:32 Orochimarufan: seems reasonable 2015-06-01 14:53:40 Orochimarufan: to fix that, i mean 2015-06-01 15:16:47 Does musl respect nsswitch.conf? 2015-06-01 15:24:42 apparently not... 2015-06-01 15:25:15 samba is pretty much useless in an AD environment when using musl >.< 2015-06-01 18:12:13 hi all, I'm trying to install apk packages without their deps. I'm looking at 'apk fetch' but not sure how to install there. Is there an equivelent to "dpkg-deb -x" on alpine? 2015-06-01 18:47:50 will untarring into root (/) be sufficient assuming no pre/post-install? 2015-06-01 18:55:09 apk fetch --quiet --stdout | tar -C / -zvx 2015-06-01 18:55:21 binate: ^^ will get you the files into the right locations 2015-06-01 18:55:34 but, obviously, doesn't do any of the processing 2015-06-01 19:00:39 tdtrask: cool, thanks 2015-06-01 19:01:01 any way to do it with processing? 2015-06-01 20:00:56 binate: it will extract the scripts needed to run, so watch the output of tar and run them 2015-06-01 20:02:39 binate: you could also change the .PKGINFO file in the package and remove the dependencies 2015-06-01 20:02:48 or change the APKBUILD and rebuild the package that way 2015-06-01 20:03:40 ahills: cool. will write a script that untars and calls any scripts if they exist... this functionality would be handy in 'apk' at some point. 2015-06-01 20:04:16 I use apk so that I don't have to manage dependencies myself... 2015-06-01 20:04:38 why don't you want the package's dependencies? 2015-06-01 20:05:01 if you want a slimmer version of something, changing the APKBUILD is definitely the way to go 2015-06-01 20:05:49 I'm building overlay images so the dependencies will be handled in a dependent image of the root fs 2015-06-01 20:06:45 Common dependencies, I should say. 2015-06-01 20:06:48 how will you handle overlays of e.g. /etc/passwd and /etc/group? 2015-06-01 20:07:08 a good portion of the scripts are just adduser'ing and addgroup'ing 2015-06-01 20:10:35 each image only runs one process so the last overlay handles those files or a postbuild script for the image. 2015-06-01 20:12:02 in general everything runs as root since the image is excuted in a container 2015-06-01 20:13:52 not "everything", the process for the container. but it could run as another user as well so the the last installed package would need to add user/group info if required. 2015-06-01 20:16:02 just means I need to peak at the pre/post install scripts to see if they do anything required for my environment. 2015-06-01 20:18:50 I see... so what do you do for dependencies that aren't common to other containers, or are only common to a few? 2015-06-01 20:20:00 i can run through a build using 'apk add' and see what deps are pulled in. 2015-06-01 20:20:41 or use apk info 2015-06-01 20:42:39 has anyone successfully used the archlinux template to create a container? 2015-06-01 20:43:16 <_ikke_> ahills: I have 2015-06-01 20:43:33 <_ikke_> you mean the archlinux image? 2015-06-01 20:44:30 yeah 2015-06-01 20:45:15 <_ikke_> I've used it, but I know use my own 2015-06-01 20:45:20 _ikke_: did you do anything more complex than installing pacman and arch-install-scripts, then lxc-create -t archlinux? 2015-06-01 20:46:00 <_ikke_> ahills: I used an existing pacman to install the base system, and then use tar + docker create to create the image 2015-06-01 20:46:05 after I configured pacman on the host, I'm now getting "mount: mount point /var/lib/lxc/archer/rootfs/dev/pts does not exist" 2015-06-01 20:46:18 ah, ok 2015-06-01 20:46:34 <_ikke_> ahills: Sorry, you are talking about lxc, I was thinkink about docker :P 2015-06-01 20:46:35 I'm not sure I want to use docker for this particular container, it doesn't fit the paradigm very well 2015-06-01 20:46:42 ohh 2015-06-01 20:47:00 well docker may be the only way to do it, this archlinux template script even has errors like missing the closing } for a ${ 2015-06-01 20:47:07 ahills: mkdir ${ROOTFS}/dev/pts -m 0755 2015-06-01 20:47:36 binate: thanks, it does look like I'm going to have to modify the template 2015-06-01 20:48:24 i haven't used lxc, but for chroot, you have to setup some stuff that similar 2015-06-01 20:49:14 I guess if it was being done from an archlinux host, it would be a simple mount --bind or similar 2015-06-01 20:49:49 I mkdir /dev, /dev/net, /dev/shm, /proc, /sys, /tmp and /dev/pts. 2015-06-01 20:50:59 ok 2015-06-01 20:52:48 you might need to mknod full, ptmx, random, urandom, zero, and tty too. 2015-06-01 20:53:14 in /dev 2015-06-01 21:03:10 ugh... 2015-06-01 21:54:28 I get the same error even when I hand-edit /usr/bin/pacstrap to create that directory 2015-06-01 22:00:27 lol these two lines, right next to each other 2015-06-01 22:00:48 /var/lib/lxc/arch0/rootfs/dev/pts 2015-06-01 22:00:48 mount: mount point /var/lib/lxc/arch0/rootfs/dev/pts does not exist 2015-06-01 23:21:25 ahills - to use non alpine lxc guests the unprivileged images from the download template http://hastebin.com/ogetakocak.rb also work 2015-06-01 23:51:13 BitL0G1c: unfortunately, I'm looking for an archlinux guest 2015-06-01 23:51:26 I did check the list of the download template 2015-06-01 23:53:31 ok 2015-06-01 23:59:29 maybe these will help https://www.archlinux.org/packages/extra/any/arch-install-scripts/ 2015-06-02 01:57:10 they're in aports 2015-06-02 01:57:27 and required for the template 2015-06-02 01:58:26 I had to modify pacstrap because udev doesn't create /dev/pts or /dev/shm (at least the way it's called) 2015-06-02 01:59:50 now pacman is complaining about the mtree specification 2015-06-02 01:59:59 which apparently is indicative of a libarchive version mismatch 2015-06-02 04:11:11 does alpine linux have selinux set to enforcing, getting permission denied for bind mounts when it doesn't make sense to 2015-06-02 07:18:54 mitchty there is no selinux, but grsec 2015-06-02 09:51:18 hello 2015-06-02 09:51:26 Is it any way to run Seamonkey? 2015-06-02 12:39:56 pLT-m3: seamonkey isn't in the package repos at the moment 2015-06-02 12:43:34 pLT-m3: Firefox is, but Thunderbird and chatzilla are not. 2015-06-02 12:44:21 thunderbird is in unmaintained though, so it could be revived 2015-06-02 14:36:46 mitchty: are you trying to escape a chroot? 2015-06-02 14:37:34 You can use the /sys filesystem to control grsec restrictions 2015-06-03 02:23:05 I'm really intrigued by Alpine, especially after I started using it as a base for Docker containers. 2015-06-03 02:24:04 I'm curious about the backend stuff. How do apks end up in the repositories? Is there a CI/CD process behind releases, or is it a maintainer based approach? 2015-06-03 02:26:57 The gatekeepers keep the gates, but I think there is CI/CD 2015-06-03 09:20:11 anyone know how do use three X displays one on each monitor? (3 screens) 2015-06-03 09:20:23 standard alpine linux 2015-06-03 09:29:17 i can stretch a desktop across three monitors with xrandr, but that's only one screen 2015-06-03 09:33:52 ScrumpyJack: can't you use a different X configs when starting them ? 2015-06-03 09:34:08 won't you have keyboard/mouse issues ? 2015-06-03 09:34:23 have no idea by the way just asking 2015-06-03 09:34:46 I've run multiple X on the same screen in the past but not on different ones... 2015-06-03 09:56:29 should probably package this https://aur.archlinux.org/packages/icecat/ 2015-06-03 09:56:37 i imagine it wouldn't be too different from the firefox apk 2015-06-03 09:58:17 coredumb: you can move the mouse across multiple screens, and windows of apps that support it 2015-06-03 09:59:29 ScrumpyJack: over multiple X ? 2015-06-03 09:59:31 nice 2015-06-03 10:01:30 but i can't get X -configure to work on a standard ivy bridge chipset 2015-06-03 10:01:42 "No devices to configure" 2015-06-03 10:49:17 when i boot a usb install of alpine linux, having added some package, they don't get installed at boot time 2015-06-03 10:49:39 <_ikke_> ScrumpyJack: You need to persist them using lbu 2015-06-03 10:49:41 i have to run apk upgrade to get them from /media/usb/cache 2015-06-03 10:49:50 yeah lbu commit done 2015-06-03 10:49:54 it's not tha 2015-06-03 10:49:57 <_ikke_> hmm 2015-06-03 10:50:22 <_ikke_> is there a symlink somewhere in /etc/apk to /media/usb/cache ? 2015-06-03 10:50:28 ah hang on, i think i removed /media/usb/cache from repositories 2015-06-03 10:51:17 i have /media/usb/apk in repositories 2015-06-03 10:51:42 yes, i have the symlink 2015-06-03 11:25:49 nice article, http://www.linuxjournal.com/content/infinite-busybox-systemd 2015-06-03 11:30:18 on page 2, "Note that even though these security precautions are taken systemd-nspawn is not suitable for secure container setups." 2015-06-03 11:30:24 ;) 2015-06-03 11:32:22 wondering why perl 6 not getting out, digging a little deeper and found it has, Native Shaped Arrays (NSA) 2015-06-03 12:36:31 I'm trying to start 4 KVM at boot time trough a /etc/local.d/kvm.start script. It does not work at boot time but woks as intended when issuing 'rc-service local restart' later on... any idea how to investigate/fix what seems to be some race condition? 2015-06-03 12:45:06 Jean-Scotch: you cannot create an initd script for them? 2015-06-03 12:45:10 like we have for lxc 2015-06-03 12:46:35 I was planning to do that later on. but first I wanted to have those KVM started automatically at boot time with "locl". Maybe I have to directly start with some lxc like scrupts... 2015-06-03 12:47:10 I suppose I'm missing something now. Maybe some ENV is not prepared yet? 2015-06-03 12:47:47 i remember i had the same issue ones, cant remember what it was though. 2015-06-03 12:48:48 oh 2015-06-03 12:49:01 Jean-Scotch: did you add local to rc? 2015-06-03 12:49:05 yes 2015-06-03 13:11:38 http://www.phoronix.com/scan.php?page=news_item&px=PyPy-2.6-Released nice! 2015-06-03 13:34:50 Jean-Scotch: what do you mean by does not work? 2015-06-03 13:38:00 ahills: the KVM are supposed to offer some services on the LAN. But a ping to they fixed IP does not respond. If I kill the VM and start it again with the same command-line (eg script), it shows herself on the network after ac ouple of seconds/ 2015-06-03 13:38:38 the network connectivity is done with openvswitch 2015-06-03 13:40:01 Oh hmm, I forget, but does local start after net, etc? 2015-06-03 13:40:06 on the baremetal to boot, only openvswitch acpid udev and local are added as rc 2015-06-03 13:40:13 local starts last in sequence 2015-06-03 13:43:53 my guess is that the OVS used to connect to the KVM is not usable yet when local is started 2015-06-03 13:45:28 I will try to replace openvswitch with plain bridges... 2015-06-03 13:47:17 you might also try to find a way to test the connectivity and wait for its availability 2015-06-03 14:04:55 I think I may have found the culprit: OVS. If lbu ci was done while some OVS where active, those are recreated at next boot and are in conflict with the ones generated when starting the KVM. 2015-06-03 14:05:11 add an cleaning line to the scripts 2015-06-03 14:05:20 and made a clean lbu ci 2015-06-03 14:05:31 add to exclusion list? 2015-06-03 14:05:56 or does it modify config while running? 2015-06-03 14:06:13 no, because it is a port on the bridge. and that is written live in the OVS DB 2015-06-03 14:07:20 is that db something that must be stored or can it be generated on start? 2015-06-03 14:08:13 the db must be stored in my case. I suppose I could rewrite a small script to reconfigure it at each boot... 2015-06-03 14:08:43 I will give a try with plain bridges. I guess the boot time will be shorter too. 2015-06-03 14:09:11 I've been meaning to set up ovs here, maybe when I get back home I will make that my next project 2015-06-03 14:09:18 manually managing ethernet bridges is a pain 2015-06-03 14:27:27 Jean-Scotch: routing the vms out on l3 should also work, with no bridges at all 2015-06-03 14:39:11 jomat: My current setup, is that the baremetal has no access to external networks. Only one KVM has. Other KVM do not have external access neither. I use 4 differents OVS for internal connectivity. 1 KVM is a firewall, 1 is a SAN, 1 is a squid+privoxy+tor and emailrelay, and last one is a X server. Baremetal just instantiates the 4 OVS then launch the 4 KVM and finally startx againt the last KVM. Those 5 machines run in run-from-r 2015-06-03 14:42:46 Jean-Scotch: you were cut off at run-from-r 2015-06-03 14:43:24 Those 5 machines run in run-from-ram mode. The SAN mounts the physical HDD as a LUKS container with LVM2 and exports the LV as NBD. 2015-06-03 14:43:29 ;) 2015-06-03 14:44:47 I like your setup, is it documented? 2015-06-03 14:45:15 <_ikke_> How can the kvms have internet access while the bare mestal doesn't? 2015-06-03 14:45:40 PCI passthrough is how I did that on Xen 2015-06-03 14:46:16 the physical NIC are added to an OVS. but this ovs is not linked at baremetal level, only in the KVM-FW 2015-06-03 14:46:31 Although I'm curious Jean-Scotch why not use the host as firewall? 2015-06-03 14:46:58 Do you not see performance degradation with the virtual interfaces? 2015-06-03 14:47:32 I want as little as possible running on the baremetal. 2015-06-03 14:47:48 Why? 2015-06-03 14:47:54 And I did not made benchmarks to compare peformances 2015-06-03 14:48:17 As an exercice of paranoid deployment ;) 2015-06-03 14:48:44 I want to insure that never a packet will going out the barebone 2015-06-03 14:49:06 Well, it will be... It still has hardware 2015-06-03 14:49:28 It is still host kernel that sends packet 2015-06-03 14:56:09 ahills: sure, but with this topology, no information is leaked about the physical NIC. And no convoluted iptables rules are needed. From the network point of view it's dead simple. 2015-06-03 15:08:04 Jean-Scotch: ah, true, you are trying to minimize trust in both directions 2015-06-03 15:29:50 Jean-Scotch: My setup is similar... I have the physical eth0 in an lxc container, which does routing for me, and which has a lot of veth-interfaces with the other ends in other vms/containers 2015-06-03 15:30:26 and of course one veth-pair between the metal and the gateway-container 2015-06-03 15:31:31 you need arp proxying that your other devices in the network know of the veth-devices in your vms 2015-06-03 15:31:47 and then there's no bridge needed at all 2015-06-03 16:33:24 I'm trying to get freeradius working on the alpine image in docker, but the radclient program always segfaults 2015-06-03 16:33:51 bad build or smth? 2015-06-03 16:34:31 running in gdb obviously doesn't make me any wiser (no debug symbols), just says "During startup program terminated with signal SIGSEGV, Segmentation fault." 2015-06-03 17:34:20 When executing setup-bootable I get: /sbin/setup-bootable: line 213: arithmetic syntax error - ? 2015-06-03 17:48:28 jamie_: on 3.2? 2015-06-03 17:48:52 yes 2015-06-03 17:51:58 jamie_: i think we changed some code in that part recently. can you add a bug to our tracker? 2015-06-03 17:53:21 explain in detail what you are trying to do, so we can check your use case. maybe also not your disk setup. 2015-06-03 17:54:09 -not :) 2015-06-03 17:54:44 jamie_: what is the size of your disk? 2015-06-03 17:58:28 unsquashfs crashes randomly on alpine 2015-06-03 17:58:44 i suspect it's stack overflows due to assuming a huge thread stack 2015-06-03 18:00:35 Trying to make a USB stick with CDROM 3.2 - USB Sandisk Cruzer 16G 2015-06-03 18:03:07 drive is formatted as instructed via alpine site 2015-06-03 18:03:56 jamie_: you want to run from ram or install it as os drive? 2015-06-03 18:07:19 checked dmesg - sr0 IO errors - is my CD drive dying? :( - LiveCD boots fine (and fast) into 3.2 ready at prompt. 2015-06-03 18:07:54 os drive 2015-06-03 18:10:25 jamie_: you can also use a second usb drive, and copy the contents of iso to usb, and afterwards install syslinux bootloader. 2015-06-03 18:14:05 only one drive. 2015-06-03 18:23:08 livecd is booting fine straight to prompt! issue setup-alpine, go through answers and get: ERROR: Failed to create lib/liblvm2cmd.so.2.02: No space left on device 2015-06-03 18:23:44 ERROR: lvm2-libs-2.02.120-r0: IO ERROR 2015-06-03 18:24:15 also used 'sys' answer to write to usb 2015-06-03 18:33:23 fyi: ERROR message randomly pops up during setup-alpine questions 2015-06-03 18:40:49 ncopa, yes unsquashfs has a stack overflow 2015-06-03 18:40:56 and it's likely exploitable 2015-06-03 18:41:20 in inflator(), char tmp[block_size]; 2015-06-03 18:41:30 block_size is a global set from contents of the file 2015-06-03 19:13:39 hope this helps - still undone - going back to wiki - thanks for your time - good day! :) 2015-06-03 19:41:09 greetings to all 2015-06-03 19:41:12 which package contains stdc-predef.h in x86_64? 2015-06-03 19:41:18 jamie_: is it possible that you are running out of space? ;) 2015-06-03 19:42:11 KoFFiE: if you want to debug, you can build from the APKBUILD, apk add alpine-sdk and git clone from aports 2015-06-03 19:43:19 vitronic: did you check pkgs.a.o? 2015-06-03 19:44:41 clandmeter, show me as 2015-06-03 19:45:11 sorry? 2015-06-03 19:45:21 i could read that in multiple ways 2015-06-03 19:47:29 vitronic, musl doesn't have stdc-predef.h yet 2015-06-03 19:47:36 i've been meaning to add it 2015-06-03 19:50:48 dalias, Strange, I've managed to compile in x86, but cannot remember if I use a trap 2015-06-03 19:53:11 what is trying to include ? 2015-06-03 19:53:19 it's not intended to be used that way 2015-06-03 19:53:29 rather gcc auto-includes it if it's available 2015-06-03 19:53:47 to make certain predefined macros whose definitions depend on libc available 2015-06-03 19:54:46 #include 2015-06-03 19:54:46 #include 2015-06-03 19:54:46 #include 2015-06-03 19:54:46 #include "barcode.h" 2015-06-03 19:55:00 only that 2015-06-03 19:55:49 have you checked barcode.h? 2015-06-03 19:56:53 ahills, if, as I said, I managed to compile X86 2015-06-03 19:58:13 I will review again barcode.h 2015-06-03 20:29:06 weird, mkinitfs for some reason isn't installing cryptsetup into my initramfs 2015-06-03 20:29:27 zcat boot/initramfs-grsec | cpio -v -t | grep cryptsetup => comes up dry 2015-06-03 20:29:45 yet... etc/mkinitfs/features.d/cryptsetup.files has /sbin/cryptsetup in it 2015-06-03 20:31:08 oh, I'm a fool, -b doesn't fix -c 2015-06-03 21:28:03 Alpine Linux vs Gentoo Hardened, thoughts? 2015-06-03 21:38:25 Apples vs Oranges? 2015-06-03 21:51:16 Green apples vs. red apples, I'd think. 2015-06-03 21:51:45 which is the master apple? 2015-06-03 21:57:47 whichever apple you choose 2015-06-03 21:58:16 kallisti 2015-06-03 22:28:48 is "JCH" here ? use lxc-console or setup ssh in the container - don't use lxc-attach 2015-06-04 00:44:00 Is the Alpine Linux live-CD image intended to be EFI-bootable? 2015-06-04 00:54:18 these 2 scripts have support for EFI http://it-offshore.co.uk/linux/alpine-linux/25-alpine-linux-luks-encrypted-installations - am going to put them up on github - you can wget them where they are now 2015-06-04 08:18:24 morning 2015-06-04 08:18:41 what's different in X11 in alpine linux? 2015-06-04 08:19:58 I can X :1 -query remotehost for XDMCP to remote host using other distros 2015-06-04 08:20:02 but not alpine linux 2015-06-04 08:20:12 to same remote host etc 2015-06-04 08:27:18 I do connect with X -query from AL to AL 2015-06-04 08:27:54 in fact I use Xorg -query 2015-06-04 08:28:36 Did you enable XDMCP in the remote AL? 2015-06-04 08:38:19 yeah, XDMCP works fine on the remote. i can access it from anywhere 2015-06-04 08:39:01 i don't know much about magic-cookies in X 2015-06-04 08:39:06 could it be that? 2015-06-04 08:44:19 error logs show "XDM too many retransmissions" 2015-06-04 08:44:37 google shows posts from <2004 2015-06-04 09:47:33 i'm getting connection refused if i try to run an X program from the remote host to alpine linux running X 2015-06-04 09:48:24 i've tried X :0 -ac on alpine linux and host + but no joy 2015-06-04 09:49:50 is grsec doing something? 2015-06-04 09:56:49 you have to pass the -listen tcp option or else ... 2015-06-04 10:38:27 now if only I could configure X with 3 displays on 3 screens so that i end up with :0.0 :0.1 and :0.2 2015-06-04 10:49:07 ScrumpyJack: might xrandr help? or is that single-screen? 2015-06-04 11:17:51 ScrumpyJack: did you allow access using xhost? 2015-06-04 12:42:43 xrandr is single screen, stretched across multiple monitors 2015-06-04 12:45:55 what does multiple screens get you? 2015-06-04 14:13:50 ahills: I have a requirement 2015-06-04 15:23:06 ncopa, did you see my messages about unsquashfs ? 2015-06-04 15:23:14 it has unbounded VLA usage :( 2015-06-04 15:59:00 apk upgrade - 2015-06-05 05:18:22 RUN gem install compass => RROR: Failed to build gem native extension.checking for ffi.h... *** extconf.rb failed ***. http://pastebin.com/Yvgqvxwp any ideas? 2015-06-05 05:18:46 my Docker file have RUN apk add --update ruby ruby-dev php-cli and RUN gem install compass 2015-06-05 06:43:36 dalias: i saw it, and then i forgot it 2015-06-05 06:45:31 dalias: care to create an issue on bugs.a.o? or send an email to alpine@bugs.alpinelinux.org with the details 2015-06-05 06:45:39 how to reproduce 2015-06-05 07:22:37 morning. the answer to yesterday's X11 three screen issue was 42 2015-06-05 07:27:32 to be more specific, xorg.conf file needed the ZaphodHead option for each Device section 2015-06-05 12:07:26 ScrumpyJack: haha 2015-06-05 12:07:58 owen1: what's ffi.h? 2015-06-05 12:12:03 http://pkgs.alpinelinux.org/contents?pkgname=libffi-dev&arch=x86 2015-06-05 12:58:20 Weird... http://pkgs.alpinelinux.org/packages does not find 'mplayer' 2015-06-05 12:59:10 indeed 2015-06-05 12:59:46 complain to the dev! 2015-06-05 13:00:07 I will do ;) 2015-06-05 13:13:23 AmatCoder: it does for x86_64 2015-06-05 13:14:39 ops 2015-06-05 13:16:03 mplayer is packaged only on x86_64 2015-06-05 13:16:14 why? 2015-06-05 13:17:45 apparently it fails to build on x86 2015-06-05 13:18:33 yeah i just saw it 2015-06-05 13:22:46 does ruby-dev need ffi.h? 2015-06-05 13:23:03 loader/ldt_keeper.o: In function `Setup_LDT_Keeper': 2015-06-05 13:23:04 /home/clandmeter/aports/main/mplayer/src/MPlayer-1.1.1/loader/ldt_keeper.c:215: undefined reference to `modify_ldt' 2015-06-05 13:23:28 AmatCoder: i'd recommen using mpv... 2015-06-05 13:23:32 instead of mplayer 2015-06-05 13:25:57 ncopa: yes, I use mpv...but I was making a APKBUILD for smplayer 2015-06-05 13:26:52 ok 2015-06-05 13:42:21 from mplayer source: 2015-06-05 13:42:23 #if defined(__GLIBC__) && (__GLIBC__ < 2 || (__GLIBC__ == 2 && __GLIBC_MINOR__ == 0)) 2015-06-05 13:42:29 _syscall3( int, modify_ldt, int, func, void *, ptr, unsigned long, bytecount ); 2015-06-05 13:42:34 #else 2015-06-05 13:42:39 int modify_ldt(int func, void *ptr, unsigned long bytecount); 2015-06-05 13:42:45 #endif 2015-06-05 13:43:19 but, why it works on x86_64? 2015-06-05 13:49:52 hi 2015-06-05 13:50:05 I have xfsprogs installed but mkfs.xfs missing? 2015-06-05 13:51:17 zxd2: on /sbin ? 2015-06-05 13:51:48 not there 2015-06-05 13:51:59 hmm 2015-06-05 13:52:08 xfsprogs-3.2 2015-06-05 13:52:20 3.2.2-r0 2015-06-05 13:52:39 just booted fresh latest alpine from iso 2015-06-05 13:52:52 and wanted to format storage with xfs 2015-06-05 13:53:54 here mkfs.xfs is on /sbin 2015-06-05 13:55:22 http://pkgs.alpinelinux.org/contents?filename=mkfs.xfs&path=&pkgname=&arch=x86_64 2015-06-05 13:55:27 should be there 2015-06-05 13:55:36 apk fix xfsprogs 2015-06-05 13:56:57 heads up 2015-06-05 13:57:02 i just pushed gcc 5.1 2015-06-05 13:57:10 and abuild 2.23.0 2015-06-05 13:57:22 abuild 2.23.0_rc2 2015-06-05 13:57:42 it will no longer pull in * as depends for -dev 2015-06-05 13:57:57 clandmeter: the arch is 32bit ! 2015-06-05 13:58:02 so the -dev packages will be more 'optimized' 2015-06-05 13:58:23 ncopa, awesome. i actually once had that in mind. good that you fixed it before i could file a bug ;) 2015-06-05 13:59:28 it will also try manage pkg-config dependencies automatically 2015-06-05 14:00:16 ncopa: what is the adv of that last part? 2015-06-05 14:01:42 file a bug? 2015-06-05 14:01:50 no mkfs.xfs in 32bit ver of xfsprogs 2015-06-05 14:02:22 zxd2: http://pkgs.alpinelinux.org/contents?filename=mkfs.xfs&path=&pkgname=&arch=x86 2015-06-05 14:02:36 there is no mkfs.ext4 either 2015-06-05 14:02:53 are you sure xfsprogs is installed? 2015-06-05 14:02:54 clandmeter: it isn't in the latest iso 2015-06-05 14:02:57 clandmeter: the idea is we can use pkgconf deeper in the development process to break cycles 2015-06-05 14:03:12 ah no it wasn't 2015-06-05 14:03:27 clandmeter: for example, makedepends="pc:audacious" is now possible 2015-06-05 14:03:39 it's confusing apk info package though it displayed installed package 2015-06-05 14:04:19 kaniini: what dpes pc: do? 2015-06-05 14:04:34 clandmeter: its a dependency type 2015-06-05 14:04:43 clandmeter: pc: == pkg-config module 2015-06-05 14:10:40 clandmeter: for now, if yo uget a warning 2015-06-05 14:11:00 if you get a warning that you need rebuild foo-dev 2015-06-05 14:11:20 then bump pkgrel 2015-06-05 14:11:36 it will likely complain of other package too 2015-06-05 14:11:51 so you recursively bump the pkgrel to the bottom 2015-06-05 14:12:57 i think after v3.3 release we can start use makedepends="pc:somelib" instead of makedepends="somelib-dev" 2015-06-05 14:13:08 not sure if we want do that though 2015-06-05 14:16:11 we are writing pkgconf related info in pkgs with abuild? 2015-06-05 14:17:02 yes 2015-06-05 14:18:20 so i guess (correct me if im wrong) its stored at build time, so pkgconf doesnt need to run multiple times afterwards? 2015-06-05 14:26:26 oh that reminds me 2015-06-05 14:26:32 abulid depends on pkgconf now 2015-06-05 14:26:51 clandmeter: no 2015-06-05 14:27:00 abuild uses pkgconf to figure out the dependencies 2015-06-05 14:27:27 pkg-config modules can have a field: Requires: 2015-06-05 14:27:48 in that case, the package shipping that module needs a 'depends=' 2015-06-05 14:28:11 or more exactly: depends="package-that-ships-" 2015-06-05 14:28:48 but to keep things a bit simpler, we let the add a provides=pc: 2015-06-05 14:29:17 then we can do: depends="pc:" without knoing exactly which package it is that provides it 2015-06-05 14:29:29 we just know that we need that pkg-config moule 2015-06-05 16:00:41 ncopa: so im going to get around to writing my open source vpn router guide for the alpinelinux wiki 2015-06-05 16:01:04 im wondering how i should split it up 2015-06-05 16:01:16 i want to have a chapter on setting up PPP 2015-06-05 16:01:27 a section on setting up routing with iptables 2015-06-05 16:01:32 and a section on subnetting with VPN 2015-06-05 16:01:44 probably should not be all in the same article, or it will be long hmm 2015-06-05 16:01:55 maybe ill have it like 2015-06-05 16:02:04 "Project Raspberry Pi VPN Router" 2015-06-05 16:02:09 and then have each going to subsections 2015-06-05 16:02:12 sub articles 2015-06-05 16:02:27 "Configuring PPP With ADSL" 2015-06-05 16:02:42 "IPtables with VPN" 2015-06-05 16:03:00 and finally a section on DNSCrypt 2015-06-05 16:11:32 * [[Linux Router with VPN]] 2015-06-05 16:11:38 i guess that's fair to say what it is 2015-06-05 17:19:12 ahills: i don't know what's ffi.h. probably a file needed when i run 'npm install compass' 2015-06-05 18:13:00 http://wiki.alpinelinux.org/wiki/Linux_Router_with_VPN_on_a_Raspberry_Pi 2015-06-05 18:13:04 coming along quite nicely :D 2015-06-05 18:13:25 might sound a bit dumb, but i think it's nice to have it all in one article 2015-06-05 18:13:44 might get people thinking about using Alpine Linux for a router that might not have thought about doing so because "too hard" 2015-06-05 19:04:01 is there a way to lock an article 2015-06-05 19:04:08 as i am working on it 2015-06-05 19:04:12 but want to go to bed :P 2015-06-05 20:01:08 ughh fricken autoconf 2015-06-05 20:20:45 hmm you know what tlsdate is missing 2015-06-05 20:20:49 a cronjob 2015-06-05 20:22:37 would that be the best place for it? 2015-06-05 20:22:50 currently alpinelinux uses ntpd part of busybox 2015-06-05 20:24:08 seems the package doesn't package the tlsdate init script either 2015-06-05 20:24:10 https://github.com/ioerror/tlsdate 2015-06-05 20:25:22 https://sources.gentoo.org/cgi-bin/viewvc.cgi/gentoo-x86/net-misc/tlsdate/files/ 2015-06-05 20:25:32 yes it seems whoever wrote the tlsdate package didn't put the openrc script in there 2015-06-05 20:27:42 ncopa: http://pkgs.alpinelinux.org/contents?pkgname=tlsdate&arch=x86_64 do you know why there's no rc script? 2015-06-05 20:30:58 might add that to the guide 2015-06-05 20:30:59 running 2015-06-05 20:31:10 openntpd on the router and updating from tlsupdate 2015-06-05 20:33:10 i might do that 2015-06-05 20:46:21 isn't openntpd a bit on the heavy side? 2015-06-05 20:46:33 yeah i decided not to use that 2015-06-05 20:46:38 ill just use regular busybox ntpd 2015-06-05 20:46:44 i just needed to add the -l option 2015-06-05 20:46:44 chrony? 2015-06-05 20:46:57 well busybox comes with an ntpd 2015-06-05 20:46:59 so ill use that 2015-06-05 20:47:59 cat /etc/conf.d/ntpd 2015-06-05 20:48:02 # By default ntpd runs as a client. Add -l to run as a server on port 123. 2015-06-05 20:48:04 NTPD_OPTS="-l -N -p " 2015-06-05 20:48:06 should be all i need 2015-06-05 20:48:12 although 2015-06-05 20:48:21 i could remove the -p 2015-06-05 20:48:30 because i don't want NTP to sync via port 123 outgoing 2015-06-05 20:48:39 ideally i'd want tlsdate to do the initial synchronization 2015-06-05 20:48:44 and then ntp to be used internally 2015-06-05 20:48:55 kind of like using dnscrypt and unbound 2015-06-05 20:49:12 though i have not tried it so i'm interested to hear from anyone who has 2015-06-05 20:49:47 https://sources.gentoo.org/cgi-bin/viewvc.cgi/gentoo-x86/net-misc/tlsdate/files/tlsdate.rc?revision=1.1&view=markup 2015-06-05 20:49:58 what did notice was the rc script seems to be missing in alpinelinux 2015-06-05 20:50:21 i'm not sure if there's a good reason for that, guess i have to ask ncopa as he packaged it 2015-06-05 23:17:29 Diftraku: ehhh openntpd isn't that heavy 2015-06-05 23:17:34 it's lighter than chrony 2015-06-05 23:17:47 compiled & lines of code 2015-06-05 23:25:58 Diftraku: I need to check but OpenNTPD devs claim to have the simplest config file 2015-06-06 01:15:11 well that's what I've heard 2015-06-06 05:18:16 ncopa: you there? 2015-06-06 05:42:15 my Docker file have RUN apk add --update ruby ruby-dev php-cli and RUN gem install compass. when I run docker-compose build I see: ERROR: Failed to build gem native extension.checking for ffi.h... *** extconf.rb failed ***. http://paste.ubuntu.com/11601852/ any ideas? 2015-06-06 05:52:18 owen1: install ffi-dev package 2015-06-06 05:55:51 kaniini: thanks! 2015-06-06 05:56:32 or its libffi-dev 2015-06-06 05:57:31 ERROR: unsatisfiable constraints: ffi-dev (missing): required by: world[ffi-dev] 2015-06-06 05:58:04 yeah try libffi-dev 2015-06-06 05:59:40 kaniini: now i am getting similar error to the first one - http://paste.ubuntu.com/11601887/ 2015-06-06 06:03:04 owen1: add also build-base pkg-config 2015-06-06 06:03:25 owen1: you are going to have to install the packages needed to build the ruby modules/ 2015-06-06 06:07:32 got it. thanks 2015-06-06 06:11:50 kaniini: i have: RUN apk add --update ruby ruby-dev libffi-dev build-base pkg-config php-cli and now i see: ERROR: unsatisfiable constraints: pkg-config (missing): required by: world[pkg-config] 2015-06-06 06:12:14 owen1: use pkgconf 2015-06-06 06:12:24 owen1: instead of pkg-config 2015-06-06 06:12:40 ok 2015-06-06 06:15:38 kaniini: works. thank you! 2015-06-06 06:49:53 hi all 2015-06-06 06:50:38 i'm trying to install elrnag but get a dependency error 2015-06-06 06:50:38 # apk add erlang 2015-06-06 06:50:38 ERROR: unsatisfiable constraints: 2015-06-06 06:50:38 erlang (missing): 2015-06-06 06:50:38 required by: world[erlang] 2015-06-06 06:50:57 would anyone know what's wrong? 2015-06-06 07:05:46 ianbytchek: Basically this is telling you that there is no such package. And there isn't in the main repository. It's in the testing repository. So my guess would be that the testing repository isn't enabled on your system. 2015-06-06 07:44:44 @teiresias thanks! true, i didn't have it enabled! 2015-06-06 08:13:00 i was able to decrypt my apk with: 2015-06-06 08:13:03 openssl enc -d -aes-256-cbc -in gateway.apkovl.tar.gz.aes-256-cbc -out gateway.apkovl.tar.gz 2015-06-06 08:13:11 but i'm trying to reverse the proceedure 2015-06-06 08:13:29 openssl aes-256-cbc -a -salt -in gateway.apkovl.tar.gz -out gateway.apkovl.tar.gz.aes-256-cbc 2015-06-06 08:13:34 doesn't seem to be the reverse 2015-06-06 08:14:38 210 $OPENSSL enc -d -$ENCRYPTION -in "$mnt/$f" | tar \ 2015-06-06 08:14:40 211 -C "$dest" -zx 2>/dev/null && return 0 2015-06-06 08:14:42 seems to be the same 2015-06-06 08:15:14 336 set -- enc "-$ENCRYPTION" -salt 2015-06-06 08:15:16 337 [ -n "$PASSWORD" ] && set -- "$@" -pass pass:"$PASSWORD" 2015-06-06 08:15:18 hmm 2015-06-06 08:16:45 seems when i try to run my decryption command 2015-06-06 08:16:52 on the package created with the encryption one 2015-06-06 08:16:54 i get 2015-06-06 08:16:56 enter aes-256-cbc decryption password: 2015-06-06 08:16:58 bad magic number 2015-06-06 08:23:29 i think i figured it out 2015-06-06 08:23:57 it looks like some sort of binary mode that it is being stored in when lbu commits it 2015-06-06 08:24:08 where as when i do it with my encryption command its armored or something in ascii 2015-06-06 08:24:29 like a neat pgp block 2015-06-06 08:25:32 ah yes, the -a option 2015-06-06 08:25:39 herr derr 2015-06-06 10:26:28 time for some food 2015-06-06 10:26:50 http://wiki.alpinelinux.org/wiki/Linux_Router_with_VPN_on_a_Raspberry_Pi 2015-06-06 10:27:11 http://wiki.alpinelinux.org/wiki/Configuring_a_Cisco_877_in_full_bridge_mode 2015-06-06 10:27:15 it's coming along quite nicely my guide 2015-06-06 12:03:41 i think it's pretty much done now 2015-06-06 12:04:06 if there was a rc script for tlsdate i would test that out and add it 2015-06-06 12:04:56 openrc scripts are really easy to do (once you've got it figured out, that is :P) 2015-06-06 12:05:22 yeah there's already one in gentoo 2015-06-06 12:05:30 im just wondering why it is not included with the tlsdate package 2015-06-06 12:05:35 when it is included with gentoo 2015-06-06 12:05:37 no clue 2015-06-06 12:05:48 maybe they added it recently or something 2015-06-06 12:06:01 well tlsdate was in testing in alpine until 3.2 2015-06-06 12:06:04 when it was moved to main 2015-06-06 12:06:42 I quite like how openrc lets you do simple scripts completely declaratively: https://github.com/Orochimarufan/custom-aports/blob/master/ntp/ntp.initd 2015-06-06 12:07:08 Yet if you really need it, you still get all the power you need to write a complex initscript 2015-06-06 12:07:43 https://sources.gentoo.org/cgi-bin/viewvc.cgi/gentoo-x86/net-misc/tlsdate/files/ 2015-06-06 12:08:00 seems to be missing the confd file too 2015-06-06 12:08:49 there looks to be two scripts and two config files, one for running tlsdate as a daemon and one for running it as a client 2015-06-06 12:08:51 maybe there's nothing to configure? :P 2015-06-06 12:08:58 been there for 2 years 2015-06-06 12:09:04 well there is 2015-06-06 12:09:06 what's tlsdate anyway? 2015-06-06 12:09:10 https://sources.gentoo.org/cgi-bin/viewvc.cgi/gentoo-x86/net-misc/tlsdate/files/tlsdate.confd?view=markup 2015-06-06 12:09:18 it looks up the date over https 2015-06-06 12:09:26 instead of ntp which does it unencrypted over 123 2015-06-06 12:09:42 *shrug* 2015-06-06 12:09:53 https://github.com/ioerror/tlsdate 2015-06-06 12:10:12 yeah i found it 2015-06-06 12:10:57 I just use ntp 2015-06-06 12:11:02 http://pkgs.alpinelinux.org/contents?pkgname=tlsdate&arch=x86_64 2015-06-06 12:11:12 yet our package seems to be missing the configuration files, not sure why 2015-06-06 12:11:33 i won't add it to the guide until it's included with the package 2015-06-06 12:11:43 guide? 2015-06-06 12:11:52 http://wiki.alpinelinux.org/wiki/Linux_Router_with_VPN_on_a_Raspberry_Pi 2015-06-06 12:12:00 feel free to comment 2015-06-06 12:12:17 i wrote that 2015-06-06 12:13:32 pi as router? I thought it had horrible ethernet :O 2015-06-06 12:13:57 well it's not really a "router" as such 2015-06-06 12:14:10 yeah, you're using it as a gateway 2015-06-06 12:14:12 but still 2015-06-06 12:14:15 a router for a home user it's fine when your internet upstream/downstream is less than 30mbit 2015-06-06 12:14:24 I guess 2015-06-06 12:14:33 course you can apply the guide to some better hardware if that suits you that's mentioned in the rationale 2015-06-06 12:14:49 We're still using a crappy oem gateway/router >.< 2015-06-06 12:14:49 a pi is just the most inexpensive, when the extra bandwidth won't make any difference 2015-06-06 12:15:12 it's still going to be a damn sight better than some d-link modem that is unmaintained and has a gazillion xss exploits 2015-06-06 12:15:25 heh 2015-06-06 12:15:42 or lets you bypass authentication altogether with setting your user agent string to a magic word 2015-06-06 12:16:02 ouch 2015-06-06 12:16:07 also d-link 2015-06-06 12:16:22 We're using AVM 2015-06-06 12:16:51 They tend to be pretty good compared to most others 2015-06-06 12:17:03 yeah except they're gpl violators if i recall 2015-06-06 12:17:15 i seem to remember reading something about them not verifying firmware either when uploading 2015-06-06 12:17:45 well, that's a double-edged sword though 2015-06-06 12:17:48 http://www.darkreading.com/vulnerabilities-and-threats/d-link-router-vulnerable-to-authentication-bypass/d/d-id/1111927 2015-06-06 12:18:09 it also allows you to upload a customized firmware image 2015-06-06 12:18:27 it's kind of the "We're all adults" mentality^^ 2015-06-06 12:19:56 the main problem with those kinds of routers is they don't really have the cpu to do crypto tunnels 2015-06-06 12:20:11 that wasn't the only one 2015-06-06 12:20:25 My pc is almost 24/7 so that's what I use for vpn 2015-06-06 12:20:30 http://www.devttys0.com/2013/10/reverse-engineering-a-d-link-backdoor/ 2015-06-06 12:21:37 http://www.insinuator.net/2014/03/how-to-own-a-router-fritzbox-avm-vulnerability-analysis/ 2015-06-06 12:23:03 there's vuns in the linux kernel too 2015-06-06 12:23:18 there's no one piece of software that doesn't have them 2015-06-06 12:23:22 but generally i think AVM is better than D-Link 2015-06-06 12:23:32 netgear is also awful 2015-06-06 12:23:39 (that goes for both commercial and open-source) 2015-06-06 12:24:03 well, yea, but a lot of them seem to be shitty XSS based ones 2015-06-06 12:24:15 the question is, is the distributing party interested in fixing those vuns 2015-06-06 12:24:18 with horrible httpds that are "rolled internally" 2015-06-06 12:24:25 usually not 2015-06-06 12:24:32 which unfortunately, a lot of companies aren't 2015-06-06 12:24:35 AVM i think is one of the better companies of fixing their shit 2015-06-06 12:24:50 D-link is terrible, and so are some of the others, eg tp-link and netgear 2015-06-06 12:26:02 i don't know why they don't just make a modem and slap openwrt on it 2015-06-06 12:26:13 luci is probably better than anything d-link could ever write 2015-06-06 12:26:15 I'd only use those for dumb hardware, like switches or cards 2015-06-06 12:26:55 even then, haven't been too happy with netgear wifi cards/sticks in the past 2015-06-06 12:27:02 yeah 2015-06-06 12:27:09 intel wireless ftw or atheros 2015-06-06 12:27:17 broadcom is badness! 2015-06-06 12:27:36 meh, I'm using a realtek wifi stick right now 2015-06-06 12:29:09 it was the only 802.11ac usb adaptor that I've found that works with linux 2015-06-06 12:29:40 yeah 802.11ac i've found is particularly awful 2015-06-06 12:29:54 what is annoying also is some manufacturers claim their thing is "openwrt ready" 2015-06-06 12:30:03 and it's not, or requires blob drivers they wont release 2015-06-06 12:30:34 in particular im thinking of the Linksys WRT1900AC 2015-06-06 12:30:40 but i think that was marvell's fault 2015-06-06 12:31:18 http://www.netis-systems.com/en/products/Dualband-Series/904.html I've found this one works rather well, even though the driver is out-of-tree which is a minor pain in the neck 2015-06-06 12:32:20 http://wiki.openwrt.org/toh/linksys/wrt1900ac looks like it is coming along 2015-06-06 12:33:11 only device i have that is 802.11ac is my phone 2015-06-06 12:33:27 i use ethernet most of the time 2015-06-06 12:33:44 my router doesn't do it either, but the situation with usb 802.11n adaptors is equally grim 2015-06-06 12:34:11 well i know the ath9k chipset supports some 802.11N equipment 2015-06-06 12:34:35 I'm pretty sure none are usb-chipsets though 2015-06-06 12:34:49 probably 2015-06-06 12:35:15 i had a broadcom in my macbook retina which i've suffered with 2015-06-06 12:35:20 its not so bad now 2015-06-06 12:35:20 and I wasn't going to buy a minipci wifi card for a 10+yo laptop that will be replaced at the next possible opportunity 2015-06-06 12:35:54 the usb one is more "universal" ;P 2015-06-06 12:35:55 oh that is a old laptop 2015-06-06 12:36:20 yeah, it came with a realtek 802.11b card 2015-06-06 12:36:47 And i decided to not suffer through that anymore^^ 2015-06-06 12:38:14 hehe that is old 2015-06-06 12:44:09 why does my ntp package pull in pkgconfig >.< 2015-06-06 13:14:01 the samba-dc package needs a dependency on tdb: 2015-06-06 13:14:03 ERROR(): uncaught exception - [Errno 2] No such file or directory 2015-06-06 13:14:12 File "/usr/lib/python2.7/site-packages/samba/tdb_util.py", line 36, in tdb_copy 2015-06-06 13:14:12 status = subprocess.call(tdbbackup_cmd, close_fds=True, shell=False) 2015-06-06 13:14:20 Orochimarufan / sillysausage: https://www.thinkpenguin.com/gnu-linux/penguin-wireless-n-usb-adapter-gnu-linux-tpe-n150usb 2015-06-06 13:14:32 I have one of these and it works very, very well 2015-06-06 13:14:38 tdbbackup is part of the tdb package 2015-06-06 13:14:44 it works just as well on my desktop as on my rpi2 2015-06-06 13:14:47 heh that's tiny 2015-06-06 13:15:16 Nice, but what's the range on it? 2015-06-06 13:15:19 so there yuou go Orochimarufan there is an atheros 802.11n that is usb 2015-06-06 13:15:53 the netis one has detachable antennas 2015-06-06 13:16:45 like this one? https://www.reddit.com/r/mildlyinteresting/comments/3851hu/this_bluetooth_dongle_has_a_fake_antenna/ 2015-06-06 13:16:50 https://www.thinkpenguin.com/gnu-linux/penguin-wireless-n-usb-adapter-w-external-antenna-gnu-linux-tpe-n150usbl that one looks more likely 2015-06-06 13:17:29 no, you can actually screw it off 2015-06-06 13:17:56 Orochimarufan: I can use it about 300ft away as I turn onto my street 2015-06-06 13:18:04 when I was using a computer as an AP 2015-06-06 13:18:10 well, it's too late now 2015-06-06 13:18:32 but I'm in an old house, so I also can't use it 30ft away in the radio deadzone that is my back stairs 2015-06-06 13:18:39 heh 2015-06-06 13:19:54 I'm fairly happy with this one anyway ;) 2015-06-07 04:27:55 i am getting erros when trying to install an bcrypt (NPM package) - http://pastebin.com/HXU8Tubb any ideas? 2015-06-07 04:28:17 i have python, make, g++, and gcc 2015-06-07 13:03:58 owen1: there's no error in that output 2015-06-07 16:57:57 hi 2015-06-07 16:58:07 I remember root was able to login via ssh without password 2015-06-07 16:58:16 how was that possible? 2015-06-07 16:58:28 want to duplicate same settings on a different distro 2015-06-07 16:59:01 <_ikke_> Why would you ever want that? 2015-06-07 16:59:15 <_ikke_> zxd: Didn't the sollution use ssh keys? 2015-06-07 16:59:30 not sure 2015-06-07 16:59:38 I think not 2015-06-07 16:59:57 <_ikke_> zxd: Allowing root to login without password is asking for trouble. 2015-06-07 17:00:10 <_ikke_> You are basically allowing everybody to do everything to that machine 2015-06-07 17:00:21 the box is isolated 2015-06-07 17:00:46 <_ikke_> The irranian nuclear servers probably too 2015-06-07 17:43:05 what's the alpine equivalent of 'build-essential'? 2015-06-07 17:43:52 <_ikke_> alpine-sdk 2015-06-07 17:44:54 <_ikke_> "abuild build-base git squashfs-tools cdrkit acct mkinitfs mtools" 2015-06-07 17:45:09 _ikke_: are u talking to me? 2015-06-07 17:45:15 <_ikke_> yes 2015-06-07 17:45:17 <_ikke_> soryr 2015-06-07 17:45:19 <_ikke_> sorry* 2015-06-07 17:45:28 so apk add alpine-sdk ? 2015-06-07 17:45:32 <_ikke_> yes 2015-06-07 17:45:34 thanks! 2015-06-07 18:02:00 i am getting some errors when 'npm install anvil-connect' - http://paste.ubuntu.com/11631075/ 2015-06-07 18:07:07 the first error i detect there is: gyp ERR! stack Error: getaddrinfo EAI_AGAIN iojs.org:443 2015-06-07 18:08:26 the iojs.org name lookup failed then 2015-06-07 18:08:42 most likely some networking/dns issue 2015-06-07 18:10:12 nsz: oh. is it something in the hands of io.js devs? 2015-06-07 18:10:19 or my laptop's networking? 2015-06-07 18:11:26 i cannot tell based on that single line 2015-06-07 18:11:39 nsz: http://paste.ubuntu.com/11631075/ 2015-06-07 18:12:52 there is only one useful line in that ouput.. 2015-06-07 18:13:13 ): 2015-06-07 18:14:28 at least you should try nslookup or strace the js nonsense that fails 2015-06-07 18:16:38 nsz: awesome! just run strace npm install -g anvil-connect ? 2015-06-07 18:16:50 how do i use nslookup? 2015-06-07 18:17:03 also, what is that useful line? 2015-06-07 18:18:52 strace nslookup iojs.org 2015-06-07 18:18:53 vs 2015-06-07 18:19:04 strace -f npm install -g anvil-connect 2015-06-07 18:20:48 fwiw here it takes a lot of time to process the results for iojs.org 2015-06-07 18:21:30 so probably there is some issue on their side in the dns config.. and npm does not handle the failures properly.. 2015-06-07 18:22:21 output of strace nslookup iojs.org - http://paste.ubuntu.com/11631578/ 2015-06-07 18:25:57 that's a horrible nslookup.. i assume it's not on alpine 2015-06-07 18:25:58 nsz: i have no issues to run the install command on my laptop. i only get that error when i am inside an alpine container - https://github.com/oren/anvil-connect/blob/master/Dockerfile 2015-06-07 18:26:16 if you see forks/clones in a strace then use -f 2015-06-07 18:26:16 nsz: yup. i run the lookup on my laptop 2015-06-07 18:26:20 otherwise it's useless 2015-06-07 18:26:55 well then you have to debug the issue from the 'docker container' 2015-06-07 18:27:09 can i instals strace on alpine? 2015-06-07 18:27:14 install 2015-06-07 18:27:24 why couldnt you? 2015-06-07 18:28:06 awesome. as long as it's available on apk i am good 2015-06-07 18:28:35 but i dont want to do hand holding debugging over irc now.. figure out why dns lookup returns with eagain 2015-06-07 18:29:24 sweet. will do. thanks! 2015-06-07 18:32:18 (inside the alpine container) strace nslookup iojs.org -> strace: test_ptrace_setoptions_for_all: PTRACE_TRACEME doesn't work: Permission denied 2015-06-07 18:32:21 (i am root) 2015-06-07 18:32:32 id -> uid=0(root) gid=0(root) groups=0(root),1(bin),2(daemon),3(sys),4(adm),6(disk),10(wheel),11(floppy),20(dialout),26(tape),27(video) 2015-06-07 18:33:44 on grsec kernel you have to do something to enable ptrace 2015-06-07 18:40:29 thanks! i'll google for it 2015-06-07 21:00:08 i tried searching for 'enable strace grsec' and also 'enable strace alpine' but couldn't find anything meaningful. anyone got good links? 2015-06-07 21:25:26 on a new installed machine "apk add freeradius" gives ERROR: unsatisfiable constraints: 2015-06-07 21:27:49 radius:~# apk add freeradius 2015-06-07 21:27:49 ERROR: unsatisfiable constraints: 2015-06-07 21:27:49 freeradius-lib-3.0.8-r3: 2015-06-07 21:27:49 conflicts: freeradius-3.0.8-r3[freeradius3=3.0.8-r3] 2015-06-07 21:27:49 satisfies: freeradius-3.0.8-r3[freeradius-lib] 2015-06-07 21:27:50 freeradius-3.0.8-r3[so:libfreeradius-eap.so] 2015-06-07 21:27:52 freeradius-3.0.8-r3[so:libfreeradius-radius.so] 2015-06-07 21:27:54 freeradius-3.0.8-r3[so:libfreeradius-server.so] 2015-06-07 21:27:56 freeradius-3.0.8-r3[freeradius-lib] 2015-06-07 21:27:58 freeradius-3.0.8-r3[so:libfreeradius-eap.so] 2015-06-07 21:28:00 freeradius-3.0.8-r3[so:libfreeradius-radius.so] 2015-06-07 21:28:02 freeradius-3.0.8-r3[so:libfreeradius-server.so] 2015-06-07 21:28:04 freeradius-3.0.8-r3[freeradius-lib] 2015-06-07 21:28:06 freeradius-3.0.8-r3[so:libfreeradius-eap.so] 2015-06-07 21:28:08 freeradius-3.0.8-r3[so:libfreeradius-radius.so] 2015-06-07 21:28:10 freeradius-3.0.8-r3[so:libfreeradius-server.so] 2015-06-07 21:28:14 freeradius-3.0.8-r3[freeradius-lib] 2015-06-07 21:28:16 freeradius-3.0.8-r3[so:libfreeradius-eap.so] 2015-06-07 21:28:18 freeradius-3.0.8-r3[so:libfreeradius-radius.so] 2015-06-07 21:28:20 freeradius-3.0.8-r3[so:libfreeradius-server.so] 2015-06-07 21:28:22 freeradius-3.0.8-r3: 2015-06-07 21:28:24 conflicts: freeradius-lib-3.0.8-r3[freeradius3=3.0.8-r3] 2015-06-07 21:28:26 satisfies: world[freeradius] 2015-06-07 22:26:05 jvl_de did you do an update of your database first? 2015-06-07 22:26:17 "# apk update" 2015-06-07 23:22:57 hi. what does apk's "unsatisfiable constraints: xyz (missing) required by [world]" mean? 2015-06-07 23:30:11 kate: It means that the package you're trying to install isn't in the repositories that you have enabled. 2015-06-07 23:33:31 kate: have you updated your repo first 2015-06-07 23:33:35 with apk update 2015-06-07 23:48:30 teiresias, i see. apk has an interesting way of phrasing that... 2015-06-07 23:48:38 systmkor, i have! 2015-06-07 23:49:59 henceforth i'll read that as "there is no package named xyz" :) 2015-06-08 00:00:32 kate: Yes, it's a bit hard to decipher. Someone else was in here asking the same question the other day. He was trying to install erlang and seeing that error message. Turns out erlang is in the testing repo, not main, and it wasn't enabled. 2015-06-08 00:01:38 i'm a programer, and i understand what satisfying things in dependency graphs involves, and treating those as predicates on graph traversal. but that still didn't relate to anything concrete for me 2015-06-08 00:12:10 kate: sorry to repeat myself about apk update, it's just sometimes I've forgotten to do it myself and wondering what's wrong and then realizing I didn't sync the database with the external mirror/repos 2015-06-08 00:12:33 kate: that said freeradius should be apart of main for 3.X 2015-06-08 00:13:25 kate: I just have 3.2 main as my repo and it comes up for me and should be installable I'll test that right now 2015-06-08 00:14:05 kate: I get the same error 2015-06-08 00:15:12 hrmm 2015-06-08 00:19:01 you didn't repeat yourself... well, not to me, at least :) 2015-06-08 00:41:52 kate: I'm going to look at the APKBUILD file for freeradius in the aports git repo 2015-06-08 00:42:42 kate: well I think someone goofed (idk for sure) 2015-06-08 00:43:11 but in the port description (i.e. the APKBUILD file) it says that freeradius-lib is a dependency but then also states that freeradius-lib is a subpackage 2015-06-08 00:43:33 soo that's a cycle in the dependency tree 2015-06-08 00:44:23 I might try changing it and see if it installs but also tempted to play video games 2015-06-08 00:47:15 and running into mysql and maria conflicts 2015-06-08 00:47:21 kate: does this need to get working today? 2015-06-08 01:07:30 i think you're confusing me with sombody else - i never mentioned freeradius 2015-06-08 01:07:45 and i don't have anything which needs resolving 2015-06-08 01:31:52 ahhhhh my bad 2015-06-08 01:32:00 sorry I thought were asking about the same thing 2015-06-08 01:34:57 i didn't see anybody ask about that :) 2015-06-08 02:03:18 quick question, how can one install older versions of apks 2015-06-08 02:03:52 i'm looking to install llvm 3.5 or some version thereof, but apk add 'llvm-dev<3.5.1' didn't work 2015-06-08 02:36:34 mitchty: it's unlikely the mirrors still carry it, but you can build it 2015-06-08 02:37:00 basically git log on the apkbuild and look for the version number you want 2015-06-08 02:37:20 you may need to also build other dependencies if they need to be older versions 2015-06-08 02:37:42 ah, k, bummer but eh 2015-06-08 02:37:58 danke ahills i'll go that route 2015-06-08 02:38:27 mitchty: if you find out that an older release of alpine has that version, it will probably still be on the mirrors 2015-06-08 02:39:10 mitchty: http://dl-3.alpinelinux.org/alpine/v3.1/main/x86_64/ 2015-06-08 02:39:28 so it looks like v3.1 had it, so take a look at your /etc/apk/repos 2015-06-08 02:39:55 http://git.alpinelinux.org/cgit/aports/commit/main/llvm/APKBUILD?id=d9a67ed0d4e40377586984b295883e767aa952cf says januaryish 2015-06-08 02:40:08 mitchty: you could add a line like @v3.1 http://mirror/v3.1/main 2015-06-08 02:40:32 I just went to that index page and searched for llvm, if 3.5.0 is a good enough version for you 2015-06-08 02:40:49 once you've added a tagged repository to /etc/apk/repositories, you can apk add llvm@tag 2015-06-08 02:40:57 e.g. apk add llvm@v3.1 2015-06-08 02:41:11 (I hope it accepts the . character, I haven't tried that exact command) 2015-06-08 02:41:14 so switching from 3.2 -> 3.1 got me 3.5 which is good enough 2015-06-08 02:41:38 for tagged repos is there a link? 2015-06-08 02:41:42 be careful when switching your whole box over 2015-06-08 02:41:55 this is to build an apk actually llvm 3.6 is too new 2015-06-08 02:42:17 mitchty: http://wiki.alpinelinux.org/wiki/Alpine_Linux_package_management#Repository_pinning 2015-06-08 02:42:21 also a docker image so no big deal there 2015-06-08 02:43:27 cool danke ahills i'll poke about with that 2015-06-08 02:44:26 bitte :) 2015-06-08 02:45:25 cross compiling alone takes ages so i got time >.< 2015-06-08 02:47:42 just annoying to hit a build error 5 hours into a compile saying that the version of opt installed is too new and even trying things don't compile 2015-06-08 02:47:51 and no make -jN doesn't work sadly 2015-06-08 02:50:27 yes... long build times make such headaches 2015-06-08 02:52:02 even worse when its a 3 stage compile for the cross compiler then again for the native 2015-06-08 02:52:29 but eh, easy enough to kick off then go for a run 2015-06-08 03:00:32 sometimes I would kick off a build before going to bed, only to find in the morning that it had failed merely five minutes after the launch 2015-06-08 03:05:42 heh yep 2015-06-08 03:05:51 takes ~6 hours to do a full build 2015-06-08 03:06:07 and lately my efforts have failed about 30 minutes in 2015-06-08 03:06:39 but its more a lack of great documentation on how to do this thats hindering me 2015-06-08 03:06:51 that and i don't think building with cross compilers is a common thing 2015-06-08 03:07:39 that said, its nice to see that musl libc is basically a non issue 2015-06-08 03:08:40 that and my by hand bit worked to a degree to get me a fully static binary that ran anywhere 2015-06-08 03:10:39 mitchty: what cross compiling are you doing? 2015-06-08 03:11:09 systmkor: compiling the glasgow haskell compiler 2015-06-08 03:11:29 i want static binaries that run anywhere 2015-06-08 03:11:39 nice 2015-06-08 03:11:39 well linux wise at least 2015-06-08 03:11:47 for what architectures 2015-06-08 03:11:54 only need x86_64 2015-06-08 03:11:56 atm 2015-06-08 03:11:58 ah 2015-06-08 03:12:10 but could do arm too if i got enough desire 2015-06-08 03:12:13 I'm planning on trying to do MIPS & ARM cross compiling for Alpine 2015-06-08 03:12:43 just annoying hitting weird linking errors relating to llvm+PIC etc.. code 2015-06-08 03:12:45 I'm thinking about hitting up the Aboriginal Linux project to see if they would be interested since they already have some level of MUSL support 2015-06-08 03:13:19 never heard of aboriginal linux 2015-06-08 03:14:02 hmm interesting 2015-06-08 03:14:08 it's from the same guy who's making toybox 2015-06-08 03:14:16 which i guess is the original creator of busybox? 2015-06-08 03:15:09 so right now i'm using the sabotage linux cross compiler stuff here https://bitbucket.org/GregorR/musl-cross 2015-06-08 03:15:32 not sure the lineage of busybox 2015-06-08 03:15:46 yah I tried that earlier and ran into issues 2015-06-08 03:15:58 but I'm supa newb for cross compiling so it was probably me 2015-06-08 03:16:33 depends, not every autoconf setup is sane 2015-06-08 03:16:53 i've had to do some wonky m4 hacking to get this to the point it works 2015-06-08 03:17:37 that and wanting llvm as a compiler backend is making this more annoying that it really need be 2015-06-08 03:19:11 but config.log is god 2015-06-08 03:20:01 so what are you looking to use aboriginal linux for then 2015-06-08 03:23:22 well it seemed like they were going for a very lightweight means of bootstrapping yourself onto other architectures 2015-06-08 03:23:42 were other things like Yocto can throw 500MB of junk you don't want 2015-06-08 03:26:45 ah 2015-06-08 03:27:07 what is the end goal use for it? embedded or routers or some other thin 2015-06-08 03:27:09 g 2015-06-08 03:28:14 it's for work 2015-06-08 03:28:23 building a custom ARM board for embedded software 2015-06-08 03:28:40 ah cool 2015-06-08 03:28:46 first things I want to try is getting Alpine to run on BeagleBone Black 2015-06-08 03:28:52 and BananaPi 2015-06-08 03:29:10 i have a radxa rock pro that works decently enough, but running debian on it atm 2015-06-08 03:29:24 and probably switching the bootloader to be u-boot 2015-06-08 03:29:46 if i could get an arm board with 4g of ram i'd be happy 2015-06-08 03:29:58 then i'd get ghc on arm 2015-06-08 03:30:10 I know there are ones with 1 or 2 gigs of ram 2015-06-08 03:30:22 I'll look around I think I may have seen one with 4G 2015-06-08 03:30:46 yep, i've got a 2g and some of the haskell source i have requires >2g of stuff to compile, i've tried :) 2015-06-08 03:30:54 that said it would be great to get an ARM board that is 64-bit ARMv8 2015-06-08 03:30:56 well for arm, for x86 its smaller 2015-06-08 03:31:28 I assume using swap space would be ungodly slow 2015-06-08 03:31:33 so what i've found for these tiny arm boards is they are massively i/o bound 2015-06-08 03:31:50 there are ARM boards with SATA2 and PCIe 2015-06-08 03:31:51 didn't work sadly, the actual ast was > 2g on arm 2015-06-08 03:32:06 I would think those are fast enough 2015-06-08 03:32:09 malloc failures all around 2015-06-08 03:32:17 ehhhh who needs malloc :P 2015-06-08 03:34:47 compilers :) 2015-06-08 03:35:36 specifically this does http://www.idris-lang.org, its a haskell package and well, the ast is huge 2015-06-08 03:35:52 fortunately it can compile to c so there is a bit of an escape hatch 2015-06-08 03:36:07 I haven't seen that language before 2015-06-08 03:36:23 its a research language so don't use it for production 2015-06-08 03:36:50 but its the most fun i've had in years programming if i'm honest 2015-06-08 03:37:29 interesting 2015-06-08 03:37:57 https://www.youtube.com/watch?v=fVBck2Zngjo 2015-06-08 03:38:08 is a good intro to a bit of why its useful 2015-06-08 03:39:01 well its what got me started down the rabbit hole 2015-06-08 03:42:46 i'll for sure watch 2015-06-08 03:43:14 well would you be down to try to get a variety of cross-compilers working for Alpine? 2015-06-08 03:47:06 for the moment if i can slay x86_64 ghc i'll be happy this week 2015-06-08 03:48:00 i understand 2015-06-08 03:48:07 :P :D 2015-06-08 03:48:30 just annoying that i can't make -jN is my biggest annoyance 2015-06-08 03:48:47 90% of the time is all one thing at a time 2015-06-08 03:56:26 i have Xorg loading the binary nvidia driver (which is linked against glibc) from within a chroot containing glibc :D 2015-06-08 03:57:16 kate: furk ya 2015-06-08 04:00:32 i wish if people insisted on releasing closed-source binaries, they would at least link them statically 2015-06-08 04:01:38 i also wish toolchains provided some mechanism to link in a library after the fact. sort of like currying for functions 2015-06-08 04:03:05 kate: would this work? 2015-06-08 04:03:05 https://nixos.org/patchelf.html 2015-06-08 04:04:34 hm, that's a nice looking tool 2015-06-08 04:05:27 it won't do what i just wished for (so i could do like: ld -thing-that-doesnt-exist nvidia_drv.so glibc.so -o nvidia_drv_without_glibc.so), but adding rpath would allow me to avoid using chroot 2015-06-08 04:12:13 kate: I remember successfully compiling that tool on Alpine but I don't think I utilized it's funcitonality yet so not 100% sure if it works 2015-06-08 04:12:36 time to grab some food 2015-06-08 04:13:34 i'll try to get this all running properly in a chroot before experimenting with the rpath idea, but i'll let you know. thanks :) 2015-06-08 04:36:46 kate: cool, let me know if I can help you out 2015-06-08 09:52:09 morning 2015-06-08 09:53:13 to add packages to a usb cache-only install, i added them to /etc/apk/world but they don't install on boot. I have to apk upgrade for them to install from the cache 2015-06-08 09:53:20 any reason why? 2015-06-08 09:57:52 they got written to /media/usb/cache when i first installed them 2015-06-08 10:16:40 hi everyone 2015-06-08 10:17:04 the openssh-askpass package seems to be broken in both latest and edge 2015-06-08 10:17:25 when I do: apk add --update openssh-askpass 2015-06-08 10:17:34 I get ERROR: unsatisfiable constraints: 2015-06-08 10:17:34 openssh-askpass (missing): 2015-06-08 10:17:35 required by: world[openssh-askpass] 2015-06-08 11:20:15 <_ikke_> c0da: That means the package cannot be found 2015-06-08 11:20:56 <_ikke_> It's in testing aparently 2015-06-08 11:20:58 <_ikke_> not in main 2015-06-08 11:35:29 ScrumpyJack: I've had that problem before too, but I don't remember what I did to make it go away... let me know if you figure it out 2015-06-08 11:35:53 it should be on a wiki page somewhere 2015-06-08 12:12:20 just tried rebeccaOS, looks nice 2015-06-08 12:12:22 http://www.phoronix.com/scan.php?page=news_item&px=Rebecca-Black-Wayland-1.8 2015-06-08 13:04:42 ScrumpyJack: re: musl not searching domains from resolv.conf, check out the suggested solution via dnsmasq at https://github.com/gliderlabs/docker-alpine/issues/8 2015-06-08 13:33:39 hey guys, can anyone help me out? I'm stuck trying to install a pip package called jsonschema and I'm getting the following error: (http://pastebin.com/dZyGGPqe) something about an SSL certificate. When I install this package using any other linux distro and pip it's fine. 2015-06-08 13:34:48 stoogenmeyer__: maybe install ca-certificates? otherwise hopefully pip has an "ignore" feature 2015-06-08 13:35:01 stoogenmeyer__: is there anything strange about your setup, like a proxy that does SSL MITM? 2015-06-08 13:36:13 not really, also tried it from different hosts so it's not the network setup 2015-06-08 13:36:46 so it's weird.. pip install jsonschema fails on that dependency (vcversioner), but if I pip install vcversioner then pip install jsonschema it works fine 2015-06-08 13:37:06 I can just add vcversioner as a manual dependency in my packages list I guess 2015-06-08 13:38:14 I don't know what sort of brain-damaged dependency problems pip has... I hate language package managers 2015-06-08 13:38:24 ahills: thanks, i'll look into that 2015-06-08 13:38:31 pip has given me many, many hours of headaches in the past 2015-06-08 13:38:33 (-: thanks anyways 2015-06-08 13:39:45 rc-update add myservice doesn't start my service on bootup, although rc-service myservice start does start my service 2015-06-08 13:39:46 can I ask a general question? What do you use linux alpine for ? 2015-06-08 13:40:15 stoogenmeyer__: xen dom0, xen domU, many servers, laptop, desktop 2015-06-08 13:40:25 same as ahills 2015-06-08 13:40:27 also containers 2015-06-08 13:40:31 yip 2015-06-08 13:40:35 so far I've only been using it for containers 2015-06-08 13:40:44 and now thin clients 2015-06-08 13:40:51 it's very small, so it's a great choice for single-purpose machines 2015-06-08 13:41:04 my laptop is alpine with alpine containers :) 2015-06-08 13:41:37 at some point I mean to have a container with glibc in it, but I haven't had to use any glibc-dependent software recently 2015-06-08 13:41:38 my *brain* runs alpine linux 2015-06-08 13:41:43 haha 2015-06-08 13:42:17 stoogenmeyer__: also using it on an rpi2 as my mpd in my house 2015-06-08 13:42:48 so you had to install a desktop environment on it? is there a dominant one that people usually use? I have a really shitty laptop and I'm considering switching to alpine 2015-06-08 13:43:20 stoogenmeyer__: 1 vps (irc, email, etc.), 1 file server (minidlna, sshfs, samba), 2x mpd on rpi 2015-06-08 13:43:34 dont use it for my laptop... yet 2015-06-08 13:43:35 ahills, ScrumpyJack: do you know how to fix these gtk menu's on alpine desktop? https://dl.dropboxusercontent.com/u/18871200/Screenshot_2015-06-04_00-59-44.png 2015-06-08 13:44:12 clandmeter: no, let me try running that client 2015-06-08 13:44:31 ncopa said it was working fine for him. 2015-06-08 13:45:00 looks fine to me too :/ 2015-06-08 13:45:13 I don't have any ~/.gtk* 2015-06-08 13:45:37 I only have ~/.config/gtk-2.0/gtkfilechooser.ini 2015-06-08 13:45:57 weird 2015-06-08 13:46:12 do you have a config I can clone? 2015-06-08 13:46:29 config for gtk? 2015-06-08 13:46:34 yes 2015-06-08 13:46:50 i never made any 2015-06-08 13:47:06 i was never an alpine desktop user, but recently i gave it a try. 2015-06-08 13:47:15 which wm? 2015-06-08 13:47:28 xfce 2015-06-08 13:48:37 4? 2015-06-08 13:48:44 ahills: i dont know if i have to do anyting regarding gtk configuration, except the settings xfce provides. 2015-06-08 13:48:52 yeah, I am going to try running xfce 2015-06-08 13:48:59 yes from 3.2 2015-06-08 13:48:59 transmission looks very different for me. probably loading gtk2 2015-06-08 13:49:11 I'm on edge, let me try pinning it to 3.2 2015-06-08 13:49:12 its not transmission 2015-06-08 13:49:20 its transmission remote 2015-06-08 13:49:24 transmission-remote-gtk, right? 2015-06-08 13:49:28 yes 2015-06-08 13:49:35 its gtk3 2015-06-08 13:50:23 in the settings, its even worse. almost impossible to read it. 2015-06-08 13:50:32 grr, I hate software that ignores -h and runs with defaults 2015-06-08 13:50:44 +1 2015-06-08 13:51:43 in this case, startx... 2015-06-08 13:55:00 ahills: i couldnt stand the fonts in alpine before. but now that i added freetype-infinality its nice. 2015-06-08 13:55:10 i cannot imagine ppl like default font rendering 2015-06-08 13:55:53 I install every font package I can find in every linux distro 2015-06-08 13:56:09 also terminus-font is a great one (thanks systmkor) 2015-06-08 13:56:27 ahills: did you add 2015-06-08 13:56:35 freetype-infinality 2015-06-08 13:57:04 it conflicts with... something 2015-06-08 13:57:13 freetype itself maybe 2015-06-08 13:57:17 oh, with freetype 2015-06-08 13:57:18 but i think its solved 2015-06-08 13:57:24 i should replace it. 2015-06-08 13:57:37 i wonder if i could add some of the patches to our regular freetype 2015-06-08 13:58:14 not sure about licensing 2015-06-08 14:00:09 what's the infinality patches? 2015-06-08 14:00:15 and why does xinit suck so much 2015-06-08 14:00:51 no matter which client I specify, it always tries to launch xterm 2015-06-08 14:01:52 you need to edit or your local or the global xiniitrc 2015-06-08 14:01:55 or whatever its called 2015-06-08 14:02:16 man xinit 2015-06-08 14:02:25 I hate old software sometimes 2015-06-08 14:03:05 ahills: https://wiki.archlinux.org/index.php/Infinality 2015-06-08 14:04:09 wow, xfce looks SO much better than when I used it in 2005 ;) 2015-06-08 14:04:22 :) 2015-06-08 14:04:23 and I don't have the issue even with that, maybe I should try infinality then 2015-06-08 14:04:42 if you want smooth fonts... 2015-06-08 14:04:59 shouldn't freetype-infinality have a replaces or provides or something? 2015-06-08 14:05:05 so that I can just apk add it? 2015-06-08 14:05:17 it has replaces 2015-06-08 14:05:22 should work 2015-06-08 14:05:26 atleast it did for me. 2015-06-08 14:05:40 it just conflicts... 2015-06-08 14:05:57 are you on edge? 2015-06-08 14:06:45 yes 2015-06-08 14:07:14 i had that issue also 2 weeks ago, but recently it resolved. 2015-06-08 14:07:25 hmmm... 2015-06-08 14:08:25 its one of the reasons i would like to move the patches to regular freetype 2015-06-08 14:08:59 ugh, I'm going to need to make a mess to install this 2015-06-08 14:09:57 shit 2015-06-08 14:10:11 i just extracted the apk last time :) 2015-06-08 14:10:19 does apk log actions? 2015-06-08 14:10:40 not that i know 2015-06-08 14:10:50 uh oh 2015-06-08 14:10:58 did you remove lots of packages by accident? 2015-06-08 14:11:00 ok, I will be back... shortly... 2015-06-08 14:11:05 yes, about 430 of them 2015-06-08 14:11:11 same here last time :) 2015-06-08 14:11:21 i added freetype again and it would add them back 2015-06-08 14:11:37 it didn't 2015-06-08 14:11:43 it added back one package 2015-06-08 14:11:49 apk upgrade? 2015-06-08 14:13:08 :) you're a wizard 2015-06-08 14:13:33 this process actually also caught an old .makedepends target 2015-06-08 14:14:15 and I can open terminals again, thank sylvanus 2015-06-08 14:16:51 clandmeter: so, I will just suggest that the package is not fixed... 2015-06-08 14:19:40 ahills: im not sure the package itself is broken. i should check with ncopa how to fix this. 2015-06-08 14:20:34 ncopa: do you have something against adding Infinality patches to freetype? 2015-06-08 14:20:57 ooh firefox fonts look much better 2015-06-08 14:21:53 i got to run. ttyt. 2015-06-08 14:22:09 ncopa: info about it here: http://bohoomil.com 2015-06-08 14:24:54 now, if only I could get firefox to be speedy... 2015-06-08 14:25:14 buy a SSD :p 2015-06-08 14:29:07 it's on one 2015-06-08 14:29:42 it was faster on arch, not sure what the difference is 2015-06-08 14:29:59 although, I rarely used it on arch, because surf didn't have as many JS errors 2015-06-08 14:30:04 with reaching the stack limit 2015-06-08 14:40:28 brb. i upgraded a router/fw and got bitten by the grsec->mvlinux-grsec/grsec.gz->initramfs-grsec name change 2015-06-08 14:50:50 re xinit 2015-06-08 14:50:58 i think there are an unresolved bug 2015-06-08 14:51:27 http://bugs.alpinelinux.org/issues/593 2015-06-08 14:53:35 undocumented, but if I give the full path to startxfce it detects it as a client, anything else only counts as an argument... 2015-06-08 14:54:11 which version of lxdm is it ? 2015-06-08 14:54:49 none? this is via xinit/startx 2015-06-08 14:55:12 oh, I didn't read the bug, sorry 2015-06-08 14:55:32 i havent checked recently 2015-06-08 14:55:53 iirc arch linux's xinit is patched 2015-06-08 14:55:56 ncopa: arch linux now has patches for xorg that run it in the same VT as startx was run, and as the user 2015-06-08 14:56:11 which would be nice to have, it's on my todo list 2015-06-08 14:56:15 yes 2015-06-08 14:56:17 that would be nice 2015-06-08 14:56:24 xenocara style 2015-06-08 14:56:28 but i'd prefer those things goes into mainline 2015-06-08 14:56:34 into upstream 2015-06-08 14:56:43 yes... but X11 upstream is very, very slow 2015-06-08 14:57:27 i thought thats why they forked x11 to xorg in the first place :) 2015-06-08 14:57:45 ok 2015-06-08 14:57:57 this fontredering thingy looks interesting 2015-06-08 14:58:12 don't try apk add freetype-infinality -f :P 2015-06-08 14:58:24 dont? 2015-06-08 14:58:34 no, it will uninstall freetype dependencies 2015-06-08 14:58:50 it is meant to replace freetype but... for some reason, the only solution is to tar xzf 2015-06-08 14:59:07 is it abi compat? 2015-06-08 14:59:10 by that I mean reverse dependencies 2015-06-08 14:59:14 yes 2015-06-08 14:59:18 well, so far it is 2015-06-08 14:59:27 clandmeter is also using it 2015-06-08 14:59:35 he might know better 2015-06-08 15:00:27 what is the reason that those patches didnt make it to upstream freetype? 2015-06-08 15:06:04 no information on the site 2015-06-08 16:36:01 speaking of X, i'm trying to create an rc service that startx X (building a thin client) 2015-06-08 16:36:26 i can do rc-service start X but it won't start at boot 2015-06-08 16:38:15 rc-update add? 2015-06-08 16:38:22 make sure you lbu commit and all that 2015-06-08 17:24:32 almost have hibernation working... 2015-06-08 17:24:41 just need to write a feature for mkinitfs 2015-06-08 18:16:28 hmm, so in order to execute a command in the initramfs, I need to add it to /usr/share/mkinitfs/initramfs-init 2015-06-08 18:16:49 but I must be reading the file wrong, because I can't seem to get the command to execute, or any output to be printed 2015-06-08 18:19:22 ncopa: ^ any advice? 2015-06-08 18:20:29 (I did zcat /boot/initramfs | cpio -i init to make sure it's the same file) 2015-06-08 18:30:10 yeah rc-update add tells me the service gets added but it doesnt start at boot 2015-06-08 18:35:57 ScrumpyJack: which run level are you adding it to? Make sure OpenRC is targeting that level? 2015-06-08 18:37:27 default 2015-06-08 18:37:43 rc-update add myservice 2015-06-08 18:40:06 it shows up in rc-update but not rc-status 2015-06-08 18:42:22 Anything weird about your init file? 2015-06-08 19:02:48 nope, i can call it using rc-service myservice start 2015-06-08 19:02:56 and it starts up fine 2015-06-08 19:36:08 does OpenRC use shell scripts to boot the system? 2015-06-08 19:36:26 can we please make a proper snmpd daemon that does not take 740kb when statically compiled 2015-06-08 19:36:39 *justopenwrtthings* 2015-06-08 19:37:06 <_ikke_> seems small to me :P 2015-06-08 19:46:09 not when you have 248kb free :3 2015-06-08 19:51:55 Diftraku: do you have any recommendations? 2015-06-08 19:53:07 ScrumpyJack: to what? 2015-06-08 19:54:36 for a smaller snmpd 2015-06-08 19:55:33 mini-snmpd with init scripts that don't suck 2015-06-08 20:00:18 this one? http://members.aon.at/linuxfreak/linux/mini_snmpd.html 2015-06-08 20:03:22 ScrumpyJack: i had a similar problem w/ openrc not detecting a service... in my case it was chrony 2015-06-08 20:03:40 i uninstalled it and then reinstalled it (and maybe rebooted) and it started working after that 2015-06-08 20:03:50 but i dont think that will help u in this case 2015-06-08 20:04:50 same symptoms: doesnt start on boot, but runs when manually executed, shows up in rc-update but not rc-status 2015-06-08 20:05:21 pnutzh4x0r: sounds familiar 2015-06-08 20:05:29 i took a look at /var/run/openrc/deptree to see if i could make sense of it 2015-06-08 20:05:30 but nothing 2015-06-08 20:10:06 you have a good memory :) 2015-06-08 20:10:37 I'm looking through the openrc docs but they are pretty sparce 2015-06-08 20:10:56 zxd - take a look at the gentoo wiki https://wiki.gentoo.org/wiki/OpenRC - in Alpine docker containers runit works very well 2015-06-08 20:22:24 yeah, i wish openrc was better documented 2015-06-08 20:22:47 i also looked at the openrc source code and tried to do some stracing 2015-06-08 20:23:22 when i did that, on the machine it wasn't working on, it appears that openrc was not seeing the file at all, even though the symlink was there in /etc/runlevels/default 2015-06-08 20:23:29 it was confusing 2015-06-08 20:23:54 all i know was that uninstalling, and reinstalling made it work *magically* 2015-06-08 20:24:11 the experience does not inspire confidence in openrc 2015-06-08 20:25:23 it's a custom init script, so there is no uninstalling/reinstalling possible :( 2015-06-08 20:26:00 i'll add some verbose logging in rc.conf (or wherever) when back at work tomorrow 2015-06-08 20:26:38 (i wonder if you need both start and stop functions. i just have a start for now) 2015-06-08 20:40:30 ScrumpyJack: you actually don't need any functions, e.g., https://github.com/ahills/beard/blob/master/beard.init.in 2015-06-08 20:48:00 you are using runscript, i thought openrc-run was the way forward 2015-06-08 20:48:41 yeah, sorry i don't have a start or stop either, i'll post tomorrow 2015-06-08 20:48:58 uh, I dunno, a commit across main/* changed things to runscript 2015-06-08 20:53:08 clandmeter: re: gtk3 2015-06-08 20:54:22 clandmeter: On xfce4, the only theme compatible with gtk3 is 'Xfce-flat' 2015-06-08 20:56:19 clandmeter: All others look bad with gtk3 apps 2015-06-08 20:57:32 clandmeter: You can select it on Menu -> Settings -> Appearance 2015-06-08 21:05:05 clandmeter: http://s3.postimg.org/7xo8yzl2b/Screenshot_2015_06_08_20_55_06.png 2015-06-08 21:33:52 albatross should look ok on xfce 2015-06-08 21:35:35 hi 2015-06-08 21:39:42 AmatCoder see ^^^ (but for spacefm the gtk2 version works best - gtk3 constantly changes & breaks stuff) 2015-06-08 21:44:12 BitL0G1c: yes, a few themes look ok with gtk2 and gtk3 apps (clearlooks, adwaita, greybird...) 2015-06-08 21:47:53 BitL0G1c: but on xfce4, only 'Xfce-flat' (from those that are installed by default) 2015-06-08 21:53:26 numix (also by the shimmer project) - is really nice 2015-06-08 21:55:40 but numix is not packaged :( 2015-06-08 21:56:22 i might do it this week - i was using albatross but am now using numix 2015-06-08 22:00:01 that would be great...numix is is widely used now 2015-06-08 22:05:39 give me an hour or so - it s/be simple to make a patch 2015-06-08 22:08:39 cool 2015-06-08 22:08:46 but no hurry 2015-06-08 22:09:01 ;) 2015-06-08 22:13:49 speaking about this: Alpine site should have a screenshoots page... 2015-06-08 22:31:13 screenshots of what? 2015-06-08 22:31:32 photos of the machines *running* alpine might be more interesting 2015-06-08 22:34:38 desktop screenshoots, of course... 2015-06-08 22:43:27 AmatCoder: alpine can have any desktop, there is no "alpine" desktop 2015-06-08 22:49:02 I know that, when I say 'desktop screenshoots' I mean 'xfce, mate, icewm, openbox, whatever...' 2015-06-08 22:50:03 AmatCoder - http://patchwork.alpinelinux.org/patch/200/ 2015-06-08 22:51:13 BitL0G1c: great! 2015-06-08 22:52:00 http://it-offshore.co.uk/images/acustom/pages/alpine-spice.jpg - alpine desktop with copy & paste functionality 2015-06-08 22:55:14 Is alpine running on qemu? 2015-06-08 22:55:44 yes that is KVM Alpine inside Debian 2015-06-08 22:56:03 http://it-offshore.co.uk/linux/alpine-linux/30-alpine-linux-spice-kvm-desktop 2015-06-08 22:56:42 oh cool 2015-06-08 22:59:34 the debian is also actually running in vmware - with vt-d you can run kvm inside a vmware linux guest - xen inside vmware doesnt work & kvm inside virtualbox also does not work 2015-06-08 23:03:51 so debian in vmware and inside alpine in kvm 8) 2015-06-08 23:04:07 which is the host? 2015-06-08 23:06:49 windoze (this is a gaming laptop) 2015-06-08 23:07:30 ops :) 2015-06-08 23:07:33 no iommu so can't virtualize windows for gaming & pass through the gpu 2015-06-08 23:08:15 i don't think any laptop mobo's support iommu unfortunately 2015-06-08 23:34:17 http://s8.postimg.org/gozb5ur11/alpine_desktop.png 2015-06-08 23:34:56 my desktop: xfce4 with clearlooks 2015-06-08 23:36:37 classical, I know... ;) 2015-06-08 23:37:57 http://s21.postimg.org/k38sv71xj/alpine_desktop_2.png 2015-06-08 23:38:24 another screenshoot with applications... 2015-06-08 23:54:05 for geany have a look at https://github.com/codebrainz/geany-themes - the vibrant ink theme is nice 2015-06-08 23:56:27 https://github.com/codebrainz/geany-themes/tree/master/screenshots 2015-06-09 00:14:17 oh nice 2015-06-09 00:15:55 I prefer light themes however 2015-06-09 00:20:42 tango-light is good 2015-06-09 00:27:33 well, I have to go...see you 2015-06-09 01:06:59 I'm just setting up alpine. Why does setup-xorg-base replace mdev with udev? What about xorg doesn't work with mdev? 2015-06-09 01:16:18 main/xorg-server is built with --enable-config-udev 2015-06-09 01:21:20 ah, ok 2015-06-09 01:22:00 (most probably done so automatic configuration of xorg works) 2015-06-09 01:22:37 with mdev you would probably need to configure xorg.conf yourself 2015-06-09 01:26:13 I keep forgetting that xorg can configure itself these days. I've been running gentoo since the xfree86 days 2015-06-09 01:26:24 and setting up X isn't something I do that often 2015-06-09 01:26:41 this may be useful https://github.com/slashbeast/mdev-like-a-boss 2015-06-09 01:27:08 thanks 2015-06-09 06:01:20 morning 2015-06-09 06:01:33 koorogi: xorg without udev does not do hotplugging 2015-06-09 06:01:50 you need udev for beeing able to plug/unplug keyboards etc 2015-06-09 06:02:40 hi ncopa 2015-06-09 06:02:57 did you see my comments about grsec in #musl ? :-p 2015-06-09 06:03:16 and also about unsquashfs VLA overflow... 2015-06-09 06:03:49 dalias, strongSwan has similar VLA overflow; though it's not as bad since the size comes from sysadmin changable config only 2015-06-09 06:04:02 neverthe less, you put wrong number in config and it crashes 2015-06-09 06:08:18 for unsquashfs it's from the file header 2015-06-09 06:08:41 and i would treat unsquashfs as dealing with untrusted files (like tar) tho perhaps it has a lot more flaws in that regard 2015-06-09 06:08:46 most practically, it simply doesn't work 2015-06-09 06:09:00 because real non-malicious files have stacks much larger than the default thread stack size 2015-06-09 06:09:08 s/stacks/VLA sizes/ 2015-06-09 06:09:23 just replacing the VLA with malloc makes it work 2015-06-09 06:15:47 dalias, you have patch? 2015-06-09 06:16:14 hi dalias 2015-06-09 06:16:18 *reading* 2015-06-09 06:16:22 not in patch form, but i can give the line # 2015-06-09 06:17:27 unsquashfs.c, inflator(), line 2102 in ver 4.3 2015-06-09 06:19:44 did you notify upstream? 2015-06-09 06:20:10 both grsecurity and squashfs-tools 2015-06-09 06:20:42 you can (could?) often find spender and pipacs in #grsecurity on OFTC 2015-06-09 06:21:18 what was the grsec issue? 2015-06-09 06:21:40 re: grsec it's just their hackish 'all-threads-setuid()' implementation 2015-06-09 06:23:02 and there was already an argument on twitter about the basic concept a while back :) 2015-06-09 06:23:12 i just noticed more issues in the implementation while reading it 2015-06-09 06:23:55 i wasn't thinking from a standpoint of it being productive to report specific issues in it, because they were already non-receptive to my views on the whole topic... 2015-06-09 06:24:35 just another "oh here's something else ugly and broken-in-corner-cases that grsec does; maybe transitioning away from grsec is a good idea.." :-) 2015-06-09 06:24:52 ok 2015-06-09 06:25:19 i have earlier thought about keeping only PaX 2015-06-09 06:26:26 iirc something like what we talked about before makes the most sense -- starting with an analysis of the features and what we're actually getting that's not in upstream kernel 2015-06-09 06:26:42 to determine if there's still significant useful stuff 2015-06-09 06:27:00 yes 2015-06-09 06:27:28 my reason for coming across this 2015-06-09 06:27:41 i'm trying to determine how to fix multi-threaded setuid correctly 2015-06-09 06:27:52 i figured 2015-06-09 06:28:01 and went to look at how they did it 2015-06-09 06:28:23 they set a flag on all tasks in the thread-group so that, on next syscall entry, they do the pending setuid 2015-06-09 06:28:45 which is rather clever but has all sorts of ugly problems they don't address well 2015-06-09 06:29:10 i figured out basically everything needed to do it right except how to stop-the-world for a thread-group 2015-06-09 06:29:51 without that, the same 'change on next syscall' approach could be done with my design, but it requires patching arch-specific code for all archs, and adds syscall overhead 2015-06-09 06:31:06 one lazy way to stop-the-world for a thread group would be invalidating the whole TLB something like munmap does but keeping the mmap_sem locked until we're done :) 2015-06-09 06:31:38 or even invalidating the page tables 2015-06-09 06:31:59 then the next instruction in all threads would TLB miss or page fault, and the handler couldn't resume execution until the mmap_sem is unlocked 2015-06-09 06:33:50 hum 2015-06-09 06:33:52 but that fails for nommu (maybe we don't care; uid races hardly matter if you don't have memory protection anyway) and it's ugly 2015-06-09 06:33:57 sounds like rather dramatic ideas :) 2015-06-09 06:34:24 i dont have much experience with kernel hacking 2015-06-09 06:34:35 it would be nice to have it fixed though 2015-06-09 06:35:04 dunno if best approach would be posting the ideas on kernel ml 2015-06-09 06:35:12 or maybe try work something out with spender 2015-06-09 06:35:31 i dont have the impression that spender is easy to work with though 2015-06-09 06:37:55 dalias: re unsquashfs, problem is: 2015-06-09 06:37:57 void *inflator(void *arg) 2015-06-09 06:37:57 { 2015-06-09 06:37:57 char tmp[block_size]; 2015-06-09 06:38:13 and that block_size is set for the squashfs image data (user input) 2015-06-09 06:38:27 right 2015-06-09 06:38:36 it should just use malloc 2015-06-09 06:38:49 that was the answer i was looking for 2015-06-09 06:38:50 thanks 2015-06-09 06:39:14 re: kernel 2015-06-09 06:39:42 i'd kinda rather go to lkml with a working prototype that can be reviewed/critiqued 2015-06-09 06:39:56 than just say "i want to do this, what's the right way?" 2015-06-09 06:40:11 *nod* 2015-06-09 06:40:21 because i have a feeling the response to the latter is going to be either ignoring it, or saying "you don't need to do this" 2015-06-09 06:40:42 seems to be the way kernel development works yes 2015-06-09 06:40:51 whereas if i have a working patch, we can get distro pressure and glibc pressure behind it 2015-06-09 06:41:01 (glibc wants a fix for this too) 2015-06-09 06:41:52 also, if you fix it properly in upstream kernel, grsecurity will have to follow (or adapt at minimum) 2015-06-09 06:41:59 :) 2015-06-09 06:42:04 well not really 2015-06-09 06:42:15 grsec is doing a hideous hack that changes the semantics of an existing syscall 2015-06-09 06:43:01 my fix would be adding new syscalls, posix_set*id or such 2015-06-09 06:43:32 that atomically change the creds of the whole process 2015-06-09 06:43:57 glibc and musl would use this if it works, and on ENOSYS fall back to the old approach 2015-06-09 06:44:16 i would probably eventually remove __synccall from musl and just have set*id() fail hard in multi-threaded processes on old kernels 2015-06-09 06:44:55 is there some way to catch that compile time? 2015-06-09 06:45:03 ? 2015-06-09 06:45:13 and surely glibc would remove their hack once the minimum supported kernel goes past the version with the new syscalls 2015-06-09 06:45:14 calling set*id() from multi-thread 2015-06-09 06:46:01 so i could get compile error "this will fail" rather than failing at runtime 2015-06-09 06:46:04 i suppose not 2015-06-09 06:46:05 glibc has this #define for "minimum required kernel for run-time"; and setting it makes it remove the unneeded fallback code 2015-06-09 06:46:13 *nod* 2015-06-09 06:46:24 it's not something i'd change right away 2015-06-09 06:46:40 first step would be fixing kernel anyways :) 2015-06-09 06:46:43 *nod* 2015-06-09 06:46:46 which does not sound trivial 2015-06-09 06:46:59 well it's not too bad 2015-06-09 06:47:09 the way i want to do it would have all the code basically isolated in one place 2015-06-09 06:47:23 in kernel/sys.c and kernel/cred.c 2015-06-09 06:57:49 unsquashfs will not have any corresponding free() 2015-06-09 07:04:45 morning all 2015-06-09 07:13:50 ncopa, it doesn't even join its threads so i don't see much of a motivation for a free here.. 2015-06-09 07:15:22 yeah... the free(tmp) would only be of symbolic value 2015-06-09 07:15:33 i was able to reproduce btw 2015-06-09 07:15:44 and the patch fixes it 2015-06-09 07:17:19 dalias: btw 2015-06-09 07:17:28 your xorg issue 2015-06-09 07:17:37 do you think you could report it on bugs.a.o? 2015-06-09 07:17:51 i might have someone who can help us fix it 2015-06-09 07:18:03 but i need the exact details and how to reproduce it 2015-06-09 07:18:14 xorg intel driver 2015-06-09 07:29:14 dalias: https://sourceforge.net/p/squashfs/bugs/59/ 2015-06-09 07:31:24 any way of making syslinux completely quiet? tried all the usual tricks but i still get the default MENU AUTOBOOT message flash up 2015-06-09 07:34:33 ncopa, well i'd have to check if it's even still reproducible 2015-06-09 07:35:04 and that probably requires making sure i know how to get back to the version of xorg i have installed now 2015-06-09 07:35:13 then testing unpinning it and upgrading 2015-06-09 07:35:46 apk add -U xorg-server<$version 2015-06-09 07:35:48 might do it 2015-06-09 07:36:01 otherwise 2015-06-09 07:36:05 boot from usb 2015-06-09 07:36:15 and try from usb 2015-06-09 07:36:55 ah yes that should work 2015-06-09 07:37:17 heh 2015-06-09 07:37:29 you could do: lbu package 2015-06-09 07:37:40 which should create an .apkovl.tar.gz file 2015-06-09 07:37:52 put that on the boot usb 2015-06-09 07:37:54 ? 2015-06-09 07:38:03 what it does: 2015-06-09 07:38:38 it will tar.gz all files in /etc/ that differ from pristine .apk provided file 2015-06-09 07:39:05 and during boot, it will extract it to tmpfs 2015-06-09 07:39:10 and run it all from tmpfs 2015-06-09 07:39:16 ah 2015-06-09 07:39:23 probably not needed tho, is it? 2015-06-09 07:39:23 which means it will try setup your currently running system 2015-06-09 07:39:43 just for conveninence, so you dont need apk add 2015-06-09 07:39:59 you should be able to just edit /etc/apk/repositories and 'apk upgrade -U -a' 2015-06-09 07:40:05 maybe if i put a blanc MENU AUTOBOOT statement ... hmm 2015-06-09 07:40:18 ScrumpyJack: try in #syslinux 2015-06-09 07:40:23 they are often helpful there 2015-06-09 07:40:23 :) 2015-06-09 07:40:28 given the amount of packages i have and how long that'd take to install, just apk add xorg or whatever should be a lot faster.. :) 2015-06-09 07:41:05 depends on how fast link you have :) 2015-06-09 07:41:17 i mean, its not apt-get... 2015-06-09 07:41:35 which might take hours 2015-06-09 07:41:54 i bet you'll resintall world in less than 5 mins 2015-06-09 07:42:18 (unless you are on dialup...) 2015-06-09 07:43:05 i had some issues with my provider, which made my connection do single thread download around 200kb 2015-06-09 07:43:16 then apk is pretty slow.. :| 2015-06-09 07:43:26 :-/ 2015-06-09 07:43:48 so now i just have a local rsynced aports on my nas :) 2015-06-09 07:44:06 i have a local repo here too 2015-06-09 07:44:18 trying to make it sync with mqtt messages 2015-06-09 07:46:54 i'm going to go to sleep now,its really late here 2015-06-09 07:47:15 but in the next day or two i'll try to do a test of latest xorg and see if it's still broken for me 2015-06-09 07:48:16 thanks 2015-06-09 07:48:24 sleep tight ;) 2015-06-09 07:48:29 we have issues with xorg+intel? 2015-06-09 07:48:38 some claim so 2015-06-09 07:48:44 i have not been able to reproduce 2015-06-09 07:48:51 what kind of issue? 2015-06-09 07:49:50 ive only applied the no-tearing fix 2015-06-09 07:50:11 http://cowboyjob.com/post/2455680/Xorg-intel-driver-was-broken-shortly-after-30-and-never-fixed 2015-06-09 07:50:40 ... 2015-06-09 07:50:46 http://cowboyjob.com/alpine-linux/2015-04-29?page=20 2015-06-09 07:51:09 this is a SEO spam bot logging irc channels in violation of channel/freenode policy 2015-06-09 07:51:40 it does look fishy indeed 2015-06-09 07:52:00 it was discussed on #musl a couple weeks ago and the bot was banned 2015-06-09 07:53:00 who is the bot? 2015-06-09 07:53:03 but at least i got a cool avatar out of it :-p 2015-06-09 07:53:09 i forget the nick 2015-06-09 07:53:23 [03:51] -!- 6 - #musl: ban *!*@158.85.162.249-static.reverse.softlayer.com [by dalias!dalias@216-12-86-13.cv.mvl.ntelos.net, 1845727 secs ago] 2015-06-09 07:54:09 could all bots in this channel please take one step forward? 2015-06-09 07:54:12 ir2ivps10 was the nick 2015-06-09 07:54:59 i think freenode glined them 2015-06-09 07:55:18 ok 2015-06-09 07:55:37 there were a bunch of bots all fitting that same nick pattern but diff #'s 2015-06-09 07:55:51 sitting on most chans on freenode :( 2015-06-09 07:57:09 wow http://cowboyjob.com is the worlds most active forum in the world! 2015-06-09 07:57:13 how do i join it? 2015-06-09 07:57:14 :) 2015-06-09 07:57:14 :D 2015-06-09 07:57:29 yeah nothing after may 19 2015-06-09 07:57:37 because that's when they got banned from freenode :) 2015-06-09 07:58:02 somebody with time to kill should send a dmca takedown to their hosting provider 2015-06-09 07:59:00 what they're doing is just putting all this irc log crap on a dummy site to get cowboyjob.com high-ranked on google 2015-06-09 07:59:09 then they're going to change it to a job listing site or something 2015-06-09 07:59:32 and it's going to be already-high-ranked for all the programming/tech search phrases 2015-06-09 08:00:19 there might also be some way to inform google and ruin their seo cred 2015-06-09 08:02:29 ncopa: regarding freetype, its just a few patches and a profile script http://git.alpinelinux.org/cgit/aports/tree/testing/freetype-infinality/APKBUILD 2015-06-09 08:03:11 clandmeter, missing h in url? :) 2015-06-09 08:03:38 clandmeter: why are those patches not upstreamed? 2015-06-09 08:04:31 ncopa: i dont know. 2015-06-09 08:04:37 dalias: missing h? 2015-06-09 08:05:13 line 7 of http://git.alpinelinux.org/cgit/aports/tree/testing/freetype-infinality/APKBUILD 2015-06-09 08:05:27 ttp:// :) 2015-06-09 08:05:31 ah yes 2015-06-09 08:05:42 i though you mean the url i pasted here :) 2015-06-09 08:06:39 i'm a little skeptical of out-of-tree patches for font rendering -- couldn't find details on what they do, and they add config file access that might be buggy 2015-06-09 08:06:45 e.g. stuff may be opened without O_CLOEXEC 2015-06-09 08:06:58 of course that might be the case in upstream freetype too :) 2015-06-09 08:07:18 fonts look soo much better with it. 2015-06-09 08:07:27 i cannot use it without it 2015-06-09 08:07:58 hurst my eyes 2015-06-09 08:08:02 it'd be nice to see some comparison pics 2015-06-09 08:08:15 rather than just font samples of mostly-ugly fonts 2015-06-09 08:08:52 dalias: i think you can find some here http://www.webupd8.org/2013/06/better-font-rendering-in-linux-with.html 2015-06-09 08:08:57 i'm interested in the patches of course 2015-06-09 08:09:10 just mildly skeptical 2015-06-09 08:13:40 i saw some screenshot 2015-06-09 08:13:59 the first screenshot does not have noticable difference 2015-06-09 08:14:10 the second one has debian looking pretty bad :) 2015-06-09 08:14:16 but that could just be debian :) 2015-06-09 08:14:23 i am definitively interested in better looking fonts 2015-06-09 08:14:36 just curious why upstream does not accept patches 2015-06-09 08:15:08 well a config file is absolutely the wrong place for this 2015-06-09 08:15:31 a freetype config file, that is 2015-06-09 08:15:45 freetype should behave deterministically based on the caller's input 2015-06-09 08:16:08 the config should go through the caller via something like fontconfig (which has awful xml config files, yes...) 2015-06-09 08:16:25 so the patches should be adding extra tunables the caller can set when calling freetype 2015-06-09 08:16:46 rather than causing freetype to load config files behind the caller's back and change its interpretation/rendering of fonts 2015-06-09 08:17:20 the patches could then add to fontconfig the ability to tune these new freetype tunables 2015-06-09 08:17:28 so you could get them globally via fontconfig 2015-06-09 08:17:31 dalias: this apk is old 2015-06-09 08:17:36 lots has changed i think 2015-06-09 08:17:42 i think we need more dalias'es 2015-06-09 08:17:59 so we could have one to work on set*uid in kernel 2015-06-09 08:18:03 :) :) :) 2015-06-09 08:18:07 and one to work on freetype 2015-06-09 08:18:09 and one to work on musl 2015-06-09 08:18:12 and one to sleep ;) 2015-06-09 08:18:21 yes that sounds like a good idea. the sleep part :) 2015-06-09 08:18:58 amazing what a good sleep can do with you mind... 2015-06-09 08:20:19 ncopa: have you tried Infinality? 2015-06-09 08:20:24 no 2015-06-09 08:20:29 i get conflict when installing 2015-06-09 08:20:34 i'd like to try it though 2015-06-09 08:20:39 or more 2015-06-09 08:20:45 i dont know why it conflicts 2015-06-09 08:20:50 its the same as regular freetype 2015-06-09 08:20:55 and has replaces 2015-06-09 08:21:01 i'd like them to get the patches upstream so when i do apk upgrade 2015-06-09 08:21:03 kabooom! 2015-06-09 08:21:09 and i have amazing fonts :) 2015-06-09 08:21:19 looks complicated to tweak config files 2015-06-09 08:21:34 i think the default is already much better 2015-06-09 08:22:36 ncopa: any idea why it conflicts? 2015-06-09 08:22:54 have 2 soltuions would also be fine for me, as long its possible to install. 2015-06-09 08:23:23 it previously worked out of the box, but just recently it started to complain. 2015-06-09 08:23:55 i'll have a look at it soonish 2015-06-09 13:20:45 ncopa: my Intel+xorg problems were fixed by enabling kms 2015-06-09 13:32:13 i would like to run a sound server in alpine linux. esd would be nice, but esd is not in the repros. is there anything else I could use? 2015-06-09 13:34:12 my clients have esd output plugins only :( 2015-06-09 13:36:27 esd shouldn't be hard to build 2015-06-09 13:39:06 to be more explicit, I have SLES9 esxi VMs that i would like to output their sounds to an alpine linux sound server. SLES9 has esd and ALSA 2015-06-09 13:46:15 ScrumpyJack: esd? 2015-06-09 13:46:20 sound system? 2015-06-09 13:46:42 is esd still maintained upstream? 2015-06-09 13:48:04 probably not 2015-06-09 13:58:17 it was superseded by pulseaudio 2015-06-09 14:01:52 i dont want introduce packages which are dead upstream 2015-06-09 14:02:33 use pulseaudio 2015-06-09 14:02:51 that's ok, i can build it privately if i need it. 2015-06-09 14:03:29 Pretty sure PA supports the esd protocol 2015-06-09 14:03:54 i thought pulseaudio was unmaintained 2015-06-09 14:03:59 (in aports) 2015-06-09 14:03:59 lol 2015-06-09 14:03:59 (If you REALLY need esd, you'd have to conjure up an APKBUILD for it. shouldn't be too hard either) 2015-06-09 14:04:12 it sort of is 2015-06-09 14:04:26 I have a working copy here https://github.com/Orochimarufan/custom-aports 2015-06-09 14:04:39 haven't upstreamed it yet 2015-06-09 14:04:41 i saw some patch for PA 2015-06-09 14:05:15 reason it went to unmaintained is it didnt build for musl and nobody cared enough at that time to fix it 2015-06-09 14:05:34 since then i have seen a patch from openembedded iirc that makes it build 2015-06-09 14:06:00 ncopa: could you show me that patch? 2015-06-09 14:06:25 ok so pulse audio could work on the alpine linux end (the sound server with the speakers) but that's not enough. the applications in the VM call /dev/dsp (the OSS i/o device) - which of course doesn't exist in the VM 2015-06-09 14:06:32 Orochimarufan: http://patchwork.openembedded.org/patch/91713/ 2015-06-09 14:06:50 ScrumpyJack: doesn't alsa come with oss compatibility? 2015-06-09 14:06:55 yes 2015-06-09 14:07:11 the oss emulation in pulseaudio is VERY broken, if that's what you're getting at 2015-06-09 14:07:23 :) 2015-06-09 14:08:08 the patch ncopa posted is to get exactly that oss emulation to compile 2015-06-09 14:08:30 the rest of pulseaudio works just fine on musl unmodified 2015-06-09 14:08:41 i'd be ok to simply patch out padsp 2015-06-09 14:08:49 padsp looked hackish at best 2015-06-09 14:09:06 that's what I did: https://github.com/Orochimarufan/custom-aports/blob/master/pulseaudio/APKBUILD#L78 2015-06-09 14:09:22 ha 2015-06-09 14:09:36 ;P 2015-06-09 14:10:35 the CFLAGS+="-D__WORDSIZE=64" is unacceptable though 2015-06-09 14:10:52 I know 2015-06-09 14:11:01 that's why it's still arch=x86_64 2015-06-09 14:11:12 I was too lazy to get a proper solution >.< 2015-06-09 14:11:16 i figured :) 2015-06-09 14:11:36 Any thoughts on a proper solution? 2015-06-09 14:11:55 If possible without patches... 2015-06-09 14:11:59 grep WORDSIZE */*.patch 2015-06-09 14:12:04 i think you need patch it 2015-06-09 14:12:14 they shouldnt use __WORDSIZE in first place 2015-06-09 14:12:36 go tell them ;P 2015-06-09 14:12:48 http://git.alpinelinux.org/cgit/aports/tree/main/drbd/bits_per_long.patch 2015-06-09 14:12:59 how it got fixed in drbd 2015-06-09 14:13:09 i think i have seen other solution for it too 2015-06-09 14:13:37 it build here with patch from oe 2015-06-09 14:14:12 and without -D__WORDSIZE... ? 2015-06-09 14:14:29 hmm let me retry 2015-06-09 14:15:02 feel free to move it back to testing/ if it builds 2015-06-09 14:15:02 https://raw.githubusercontent.com/openembedded/oe-core/master/meta/recipes-multimedia/pulseaudio/pulseaudio/0001-padsp-Make-it-compile-on-musl.patch 2015-06-09 14:15:20 i didnt use the one from unmaintained 2015-06-09 14:15:41 clandmeter: other alternative seems to be --disable-oss-wrapper https://github.com/Orochimarufan/custom-aports/blob/master/pulseaudio/APKBUILD#L78 2015-06-09 14:16:05 push something that builds to testing/ and remove it from unmaintained 2015-06-09 14:17:37 it builds alright, but: 2015-06-09 14:17:41 pulsecore/sample-util.h:58:5: warning: "__WORDSIZE" is not defined [-Wundef] 2015-06-09 14:17:41 #if __WORDSIZE == 64 || ((ULONG_MAX) > (UINT_MAX)) 2015-06-09 14:18:20 I guess you could call it dirty programming *shrug* 2015-06-09 14:18:58 #if ((ULONG_MAX) > (UINT_MAX)) 2015-06-09 14:19:01 should work 2015-06-09 14:19:17 right, it just spews around warnings 2015-06-09 14:19:28 yes, only warnings 2015-06-09 14:19:56 i can push my very basic apkbuild. 2015-06-09 14:20:00 all right, mystery solved 2015-06-09 14:20:40 hmm, should there be a pulseaudio-clients package or something? for pacat pactl etc? 2015-06-09 14:21:19 i was going over the one in unmaintained, but it doesnt make much sence 2015-06-09 14:21:27 but im no pa expert 2015-06-09 14:22:25 i guess we can split the libs 2015-06-09 14:24:54 would it be worthwile to patch out the __WORDSIZE warnings? 2015-06-09 14:25:58 >>> WARNING: pulseaudio-dev*: glib-dev should be removed from depends_dev 2015-06-09 14:26:02 what's that mean? 2015-06-09 14:26:49 i think you can move it to depends 2015-06-09 14:26:59 recent changes in abuild detects them. 2015-06-09 14:27:07 but ncopa can explain it better. 2015-06-09 14:27:08 well, but what's the reason 2015-06-09 14:27:22 I'd like to actually understand what I'm doing^^ 2015-06-09 14:27:49 read backlog or irclogger 2015-06-09 14:27:59 1-2 days ago it was explained 2015-06-09 14:28:10 let me see 2015-06-09 14:30:07 ACTION sighs 2015-06-09 14:30:21 ACTION adds "write a replacement pulseaudio" to the ever-growing todo list 2015-06-09 14:30:42 heh 2015-06-09 14:30:46 good luck 2015-06-09 14:31:08 luck is not needed really, i just need to add protocol support to openbsd's sndio stuff 2015-06-09 14:32:28 I'm looking forward to it ;) 2015-06-09 14:33:05 so, according to http://dup.pw/abuild/9e917553 it says that for pc dependencies 2015-06-09 14:33:25 Very insightful... 2015-06-09 14:36:18 can you override arch in subpackages? 2015-06-09 14:37:36 or rather, should it be done (for -doc subpackages)? 2015-06-09 14:37:46 (since: >>> WARNING: pulseaudio-doc*: No arch specific binaries found so arch should probably be set to "noarch") 2015-06-09 14:38:44 Orochimarufan: are you writing your own -doc function? 2015-06-09 14:38:51 yes 2015-06-09 14:38:57 should I not be? 2015-06-09 14:39:04 depends 2015-06-09 14:39:10 i dont think so in PA case 2015-06-09 14:39:20 autocrap will handle it 2015-06-09 14:39:21 teach me how to do it properly then ;) 2015-06-09 14:39:31 and abuild will auto create it 2015-06-09 14:39:53 will it? 2015-06-09 14:39:57 so what do I do? 2015-06-09 14:40:10 for instance subpackages="$pkgname-dev $pkgname-doc $pkgname-libs" 2015-06-09 14:40:21 only the libs is a custom fucntion 2015-06-09 14:40:30 lets see... 2015-06-09 14:40:38 -dev and -doc will be detected by abuild and do magic 2015-06-09 14:41:31 ACTION tries that 2015-06-09 14:42:12 Orochimarufan: PA is in aports now 2015-06-09 14:42:41 im adding the openrc scripts from unmaintained (dont now if they work) and move libs to seperate pkg 2015-06-09 14:42:51 I'm considering submitting my version though, since you said it was very basic 2015-06-09 14:43:03 i added it 10 min ago 2015-06-09 14:43:05 :) 2015-06-09 14:43:25 doesn't mean it can't be superseded, he? :P 2015-06-09 14:43:38 It's still in testing, after all 2015-06-09 14:43:41 you can submit patches, which we prefer. 2015-06-09 14:43:59 mine is really basic PA 2015-06-09 14:44:12 cause i dont know what everybody wants in it. 2015-06-09 14:44:20 some dont even want it at all :) 2015-06-09 14:44:53 ^^ 2015-06-09 14:47:37 Orochimarufan: does plex server work on alpine? 2015-06-09 14:47:44 in a way 2015-06-09 14:47:56 you won't like what I did to make it ;) 2015-06-09 14:48:18 you patched it 2015-06-09 14:48:50 anyway, fixed pa APKBUILD: https://github.com/Orochimarufan/custom-aports/blob/master/pulseaudio/APKBUILD 2015-06-09 14:50:04 though I'll probably put some more work into it 2015-06-09 14:50:04 Orochimarufan: you made it work with patchelf? and it doesnt crash everyday? 2015-06-09 14:50:32 nope, works like a charm 2015-06-09 14:50:39 the key is my glibc-runtime package 2015-06-09 14:51:02 it's a HACK that deserves to be called so :P 2015-06-09 14:51:24 is glibc runtime really needed? 2015-06-09 14:51:29 it is 2015-06-09 14:51:34 did you try recently? 2015-06-09 14:51:41 unless you want to reverse-engineer pms 2015-06-09 14:51:44 we added glibc compat 2015-06-09 14:51:50 where? 2015-06-09 14:51:53 in musl 2015-06-09 14:52:02 link or something? 2015-06-09 14:52:08 well musl should be somewhat abi compat 2015-06-09 14:52:27 I might retry at some point 2015-06-09 14:52:28 check one of the latest commits in musl 2015-06-09 14:53:13 we made flashplayer work :) 2015-06-09 14:53:48 neat 2015-06-09 14:54:01 I don't use Alpine for my desktop though :) 2015-06-09 14:55:08 hmm I don't know what commit you're referring to -.- 2015-06-09 14:58:30 Orochimarufan: http://git.alpinelinux.org/cgit/aports/commit/main/musl?id=1774dd06afa61cf1f8da3cb4ab1033becc889ad0 2015-06-09 14:59:14 hmm I see 2015-06-09 14:59:17 let me try 2015-06-09 15:01:19 btw ncopa might I ask you to get something like this into abuild: https://github.com/Orochimarufan/custom-aports/blob/master/plex-media-server/APKBUILD#L21 ? 2015-06-09 15:02:16 Orochimarufan: re glib-dev removed from depends_dev 2015-06-09 15:02:35 abuild will track pkg-config depends for you 2015-06-09 15:02:46 so you can move glib-dev to makedepends 2015-06-09 15:02:59 I thought something along those lines 2015-06-09 15:03:10 and abuild will automatically handle the pkg-config depends 2015-06-09 15:03:43 I still don't get the difference between depends_dev and makedepends really :O 2015-06-09 15:04:12 depends_dev = $pkgname-dev's depend 2015-06-09 15:04:23 ah 2015-06-09 15:04:26 that makes sense 2015-06-09 15:04:35 its a hack 2015-06-09 15:04:51 which needs a better, general solution 2015-06-09 15:05:03 it's for the -dev magic^^ 2015-06-09 15:05:24 that I didn't even know existed till today :O 2015-06-09 15:06:11 anyway, about the arch-dependent sources? 2015-06-09 15:06:45 hum 2015-06-09 15:06:52 i dont like 2015-06-09 15:06:58 why not? 2015-06-09 15:07:14 right now it's a serious mess 2015-06-09 15:07:33 and as you can probably tell, my attempts at fixing the mess failed pretty hard 2015-06-09 15:07:55 hum 2015-06-09 15:08:07 i have no good idea 2015-06-09 15:08:11 maybe: 2015-06-09 15:08:20 I believe this is how Arch does it in their PKGBUILDs 2015-06-09 15:08:43 source="http://...$_debarch.deb" 2015-06-09 15:08:58 case "$CARCH" in 2015-06-09 15:09:07 x86) _debarch=i386;; 2015-06-09 15:09:19 x86_64) _debarch=amd64;; 2015-06-09 15:09:20 no, it needs to be declarative for "abuild checksum" to pick up all of them 2015-06-09 15:09:20 esac 2015-06-09 15:09:29 oh 2015-06-09 15:09:30 true 2015-06-09 15:09:48 look at the add_arch part at the end of that file 2015-06-09 15:10:11 (I don't in any way shape or form claim that mine is a good solution^^) 2015-06-09 15:10:20 whoops i need to run 2015-06-09 15:10:25 i will think about it 2015-06-09 15:10:27 but generally 2015-06-09 15:10:38 something like this is a good reason for change the APKBUILD format 2015-06-09 15:10:45 see u 2015-06-09 15:11:03 the question is, change it into what? cu 2015-06-09 15:11:05 overide default_unpack? 2015-06-09 15:11:39 whatever it is, it should work with the workflow (which included abuild checksum) 2015-06-09 15:11:51 s/included/includes/ 2015-06-09 15:14:02 btw clandmeter, you seem to have forgotten to remove pulseaudio from unmaintained/ 2015-06-09 15:17:10 i didnt 2015-06-09 15:17:12 also clandmeter, plex doesn't work with libc6-compat: 2015-06-09 15:17:24 Error relocating ./libxml2.so.2: __isinf: symbol not found 2015-06-09 15:17:24 Error relocating ./libavutil.so.54: __isnan: symbol not found 2015-06-09 15:17:24 Error relocating ./libavutil.so.54: __isinf: symbol not found 2015-06-09 15:17:24 Error relocating ./libavcodec.so.56: __isnanf: symbol not found 2015-06-09 15:17:25 Error relocating ./libavcodec.so.56: __isinff: symbol not found 2015-06-09 15:17:26 Error relocating ./libavcodec.so.56: __isnan: symbol not found 2015-06-09 15:17:27 Error relocating ./libavformat.so.56: __isnan: symbol not found 2015-06-09 15:17:28 Error relocating ./libcurl.so.4: __strtok_r: symbol not found 2015-06-09 15:17:29 Error relocating ./libcurl.so.4: __rawmemchr: symbol not found 2015-06-09 15:17:30 Error relocating ./libpython2.7.so.1.0: __finite: symbol not found 2015-06-09 15:17:31 Error relocating ./libpython2.7.so.1.0: __rawmemchr: symbol not found 2015-06-09 15:17:33 Error relocating ./libpython2.7.so.1.0: __isinff: symbol not found 2015-06-09 15:17:34 Error relocating ./libpython2.7.so.1.0: tmpnam_r: symbol not found 2015-06-09 15:17:35 Error relocating ./libpython2.7.so.1.0: __isnan: symbol not found 2015-06-09 15:17:35 :/ 2015-06-09 15:17:36 Error relocating ./libpython2.7.so.1.0: __isinf: symbol not found 2015-06-09 15:17:37 Error relocating ./libgnsdk_dsp.so.3.07.5: __strtod_internal: symbol not found 2015-06-09 15:17:38 Error relocating ./libgnsdk_manager.so.3.07.5: backtrace_symbols_fd: symbol not found 2015-06-09 15:17:39 Error relocating ./libgnsdk_manager.so.3.07.5: backtrace: symbol not found 2015-06-09 15:17:40 Error relocating ./libjemalloc.so.1: __register_atfork: symbol not found 2015-06-09 15:17:41 Error relocating Plex Media Server: getcontext: symbol not found 2015-06-09 15:17:53 (sorry for the wall of text -.-) 2015-06-09 15:18:59 i kept unmaintained for now for reference. ill remove it later 2015-06-09 15:19:05 I see 2015-06-09 15:19:26 Nothing wrong with that (although that's what git history is for, no?) 2015-06-09 15:19:35 do you need any special option enabled on PA? 2015-06-09 15:19:44 I don't think so 2015-06-09 15:20:06 I'm using a specially patched version myself anyway 2015-06-09 15:20:28 And that's a patch that shouldn't go upstream 2015-06-09 15:33:14 btw, /usr/lib/pulseaudio shouldn't go into -libs 2015-06-09 15:33:32 it should stay in the main package, because it contains only the server modules 2015-06-09 15:34:21 the -libs are as libpulse*.so like this: https://github.com/Orochimarufan/custom-aports/blob/master/pulseaudio/APKBUILD#L120 2015-06-09 15:34:58 actually, I'm contradicting myself right now 2015-06-09 15:36:51 right... /usr/lib/pulse-x.x has the modules, /usr/lib/pulseaudio is actually correct 2015-06-09 15:37:43 does -libs do magic for /usr/lib/*.so or are you missing the libpulse*.so nonetheless? 2015-06-09 15:45:14 Orochimarufan: some changes now into PA 2015-06-09 15:45:21 but still needs work 2015-06-09 15:45:40 need to split libs into seperate pkgs 2015-06-09 15:45:53 ? 2015-06-09 15:47:16 some are modules and are linked inviddually, so to keep deps low on libs we can split them. 2015-06-09 15:47:25 ah i see 2015-06-09 15:47:31 I think arch did that recently 2015-06-09 15:47:39 http://pkgs.fedoraproject.org/cgit/pulseaudio.git/tree/pulseaudio.spec 2015-06-09 15:48:00 going home. ciao 2015-06-09 15:48:04 cu 2015-06-09 15:49:47 https://projects.archlinux.org/svntogit/packages.git/tree/pulseaudio/trunk/PKGBUILD#n98 2015-06-09 18:27:08 why the F does pulse link everything against xcb >.< 2015-06-09 18:35:14 -/var/build/x2/usr/lib/pulse-6.0/modules/libprotocol-http.so.txt: libX11-xcb.so.1 2015-06-09 18:35:15 -/var/build/x2/usr/lib/pulse-6.0/modules/libprotocol-http.so.txt: libX11.so.6 2015-06-09 18:35:24 In which world does HTTP require X11... 2015-06-09 18:35:36 because it's freedesktop crap which should be avoided 2015-06-09 18:50:40 stf: is there another audio-over-tcp solution right now? 2015-06-09 18:58:06 is latency an issue? 2015-06-09 18:58:28 if it is, pulse doesn't do a too pretty job anyway^^ 2015-06-09 18:58:31 i never had the need for this, http://jackaudio.org/faq/netjack.html 2015-06-09 18:59:34 i convert my audio to wav and use tcp, but i use this only for ham radio of digital signals, for decoding on another host. 2015-06-09 18:59:53 netcat :P 2015-06-09 19:02:07 Anyway, this is the best I could come up with for Pulseaudio: https://github.com/Orochimarufan/custom-aports/blob/master/pulseaudio/APKBUILD 2015-06-09 19:17:46 how can i remove a package which some other package depends on? 'apk del -f openrc' seems to have no effect 2015-06-09 19:24:50 --yes-i-actually-mean-it 2015-06-09 19:40:12 aha, -d 2015-06-09 19:45:55 no, that was for something else. i'm confused :( 2015-06-09 20:02:09 kate: i dont think you can 2015-06-09 20:03:09 i think openrc is pulled in by alpine-base or something similar 2015-06-09 20:03:24 apk wont let you remove it, not even by force 2015-06-09 20:08:31 well maybe apk and i have different ideas of what "force" means 2015-06-09 20:11:18 kate: i guess you are trying another init? 2015-06-09 20:14:33 clandmeter: I finished splitting up PA, if you're interested: https://github.com/Orochimarufan/custom-aports/blob/master/pulseaudio/APKBUILD 2015-06-09 20:59:22 clandmeter, no, the default (busybox's) /sbin/init. but i have its /etc/inittab set to run something other than openrc's rc stuff. just a straight-up shell script, actually 2015-06-09 21:00:37 so i just wanted to remove openrc to get rid of all the distracting stuff in /etc, so i can more easily see what i'm actually using 2015-06-09 21:04:17 kate: thats what i mean with init, openrc is an init system :) 2015-06-09 21:05:05 confusing terminology :( init(8) ought to be termed an init system, imo... 2015-06-09 21:05:24 Orochimarufan: i will be commiting my changes. probably similar to what you have. 2015-06-09 21:05:51 clandmeter: Not like I have any say in the matter ;) 2015-06-09 21:06:00 kate: yes it is sometimes 2015-06-09 21:12:41 kate: you could build your own version of http://git.alpinelinux.org/cgit/aports/tree/main/alpine-base/APKBUILD without openrc support. 2015-06-09 21:15:35 but that doenst prevent other packages from installing their init scripts 2015-06-09 21:15:45 You don't need alpine-base do you? 2015-06-09 21:16:19 You could just cherry-pick packages from the list 2015-06-09 21:16:21 its a meta package 2015-06-09 21:16:44 Does anything explicitly depend on it? 2015-06-09 21:17:28 i never tried it :) 2015-06-09 21:17:47 you could install all others manually which will add them to world 2015-06-09 21:18:52 on tmpfs it will be hard, cause init will reinstall it i guess. disk install could be possible. 2015-06-09 21:20:05 anyway, calling it a day, cu 2015-06-09 21:20:49 bye 2015-06-09 22:39:21 ahills, that's what i'm doing :) 2015-06-10 06:20:01 anyone using weechat on alpine? 2015-06-10 06:20:22 when i join the channel the screen looks like: http://i.imgur.com/Pw8SVfd.png 2015-06-10 06:20:33 if you think there's an issue related to musl, might ask on #musl 2015-06-10 06:20:37 iirc some ppl in #musl use it 2015-06-10 06:20:57 i'm not sure its musl or if its xfce4-terminal 2015-06-10 07:33:23 Hello. 2015-06-10 07:33:37 algitbot: hello there, little robot 2015-06-10 07:34:17 I've decided to switch from Kali Linux to something else and currently deciding between Alpine and Void 2015-06-10 07:35:18 Had a question about the APKs being stored in RAM: if I get to build a big package (e.g. Blender) will I be able to run it without wasting time on storing the installed files in RAM? 2015-06-10 07:36:04 I have 8GB on my netbook, so that's plently of space... but since I dedicate some of it to tmpfs on /tmp, I won't have much left with lots of tabs open 2015-06-10 07:40:02 flatroze, you can do regular hard disk install too no problem 2015-06-10 07:40:05 that's how i run my laptop 2015-06-10 07:40:45 Alpine and Void are relatively similar. i think we have lot in common. but there are distinct areas where other is more mature. 2015-06-10 07:41:53 "relatively similar" meaning in ideological level. there's big differences like package manager. 2015-06-10 07:54:24 thanks fabled 2015-06-10 07:54:59 that's why I've ended up with those two choices, I believe... they both use runit, too, if I'm not mistaken? 2015-06-10 07:58:28 no, we use openrc at the moment 2015-06-10 07:58:53 but yeah, i think both want to stay rid of systemd 2015-06-10 08:08:19 we do have runit packaged though 2015-06-10 08:11:39 systemd is one of the reasons I'm leaving the mainstream linux... it's also not what it used to be, feels more like the WindowsXP crowd has moved to Ubuntu/Debian/SUSE nowadays and it's giving things like systemd a way to go :( 2015-06-10 08:12:46 I thought about going the *BSD way but those folks don't have enough stable drivers for me to run the system on a relatively new laptop, plus they don't have KVM and therefore I won't be able to virtualize things properly in there 2015-06-10 08:13:19 so GNU/Linux is still what I want to use, just don't like what it has turned into, comparing to what it was 7-8 years ago when I switched to it from windows 2015-06-10 08:29:59 flatroze: you have bhyve at least on FreeBSD, which is stable 2015-06-10 09:10:07 for alpine desktop users, i just pushed new freetype with infinality to edge. I will be removing freetype-infinality from testing. 2015-06-10 09:21:26 Hi all, we at LeaseWeb have decided to mirror Alpine Linux. Where do i announce/report new mirrors? 2015-06-10 09:28:12 nice! 2015-06-10 09:30:14 seems like you can use them, because rsyncing took a while :D 2015-06-10 09:30:37 hrak: someone else here will surely now. alternatively, if that doesn't work there's a infra bugtracker: https://bugs.alpinelinux.org/projects/alpine-infra 2015-06-10 09:30:41 *know 2015-06-10 09:31:14 flatroze: Welcome to Alpine Linux! :) 2015-06-10 09:32:16 hrak: mooi :) 2015-06-10 09:33:01 we are only very slightly GNU/Linux, as we don't use the GNU C libraries. We do compile using GCC :( but that may change 2015-06-10 09:34:20 darkfader: ok i'll open a ticket there with a list 2015-06-10 09:36:04 hrak: did you sync from nl.a.o? 2015-06-10 09:36:48 ScrumpyJack: for the gnuness of gnu/linux i think the userspace implementation of posix is what matters (shell utilities and libc interfaces) not the compiler you used to build the system 2015-06-10 09:37:25 clandmeter: i used the one from the mirror instructions (rsync.alpinelinux.org) 2015-06-10 09:37:47 yes its nl.a.o if im correct 2015-06-10 09:37:53 shouldnt be too slow though 2015-06-10 09:38:35 but a fast mirror is always welcome. 2015-06-10 09:39:12 i set up our NL mirror to sync with that one, and our Frankfurt, US eastcoast and US westcoast mirror sync from our NL mirror 2015-06-10 09:39:30 hrak: what is the url? i can commit it right away. 2015-06-10 09:39:52 NL is currently 4Gbit but will soon be 10Gbit, rest is all 10G 2015-06-10 09:41:12 i guess http://mirror.nl.leaseweb.net/alpine 2015-06-10 09:42:16 nsz: not to me :) for me it's also the build tool(chain) :( 2015-06-10 09:47:04 clandmeter: we just created a new DNS record for the NL mirror which should be active within an hour. When that is active it will look like this: 2015-06-10 09:47:23 NL= http://mirror.ams1.nl.leaseweb.net/alpine 2015-06-10 09:47:40 DE= http://mirror.fra10.de.leaseweb.net/alpine 2015-06-10 09:47:59 US east coast= http://mirror.wdc1.us.leaseweb.net/alpine 2015-06-10 09:48:20 US west coast (not synced yet)= http://mirror.sfo12.us.leaseweb.net/alpine 2015-06-10 09:49:46 there is also a GeoDNS enabled url http://mirror.leaseweb.com/alpine which should direct you to the nearest mirror. 2015-06-10 09:50:22 or mirror.leaseweb.net for that matter 2015-06-10 09:56:51 thx hrak 2015-06-10 09:56:54 gr8 2015-06-10 10:46:04 hrak: are you ok with adding just http://mirror.leaseweb.com/alpine ? 2015-06-10 12:07:51 whee 2015-06-10 12:08:00 i got a font revolution here 2015-06-10 12:08:13 font rendering revolution 2015-06-10 12:12:24 :) 2015-06-10 12:19:13 hum 2015-06-10 12:19:28 seems "sans" bold font is no longer sans 2015-06-10 12:21:37 http://i.imgur.com/uZXAG3y.png 2015-06-10 12:21:53 is that correct look of Sans, Bold? 2015-06-10 12:23:57 hum 2015-06-10 12:24:09 The "serif" and "sans" are identical 2015-06-10 12:27:04 i have no machine to test now. 2015-06-10 13:03:23 hi 2015-06-10 13:03:36 freetype update breaks monospace fonts 2015-06-10 13:03:57 http://s29.postimg.org/ldpqg2irr/Screenshot_2015_06_10_12_53_19.png 2015-06-10 13:05:02 it is the default font on xfce4-terminal 2015-06-10 13:07:38 AmatCoder: is it related to freetype of fontconfig? 2015-06-10 13:08:03 I think so 2015-06-10 13:08:16 s/of/or 2015-06-10 13:09:06 freetype, I guess 2015-06-10 13:10:07 I am not sure, I just make 'apk update' 2015-06-10 13:10:24 both freetype and fontconfig were upgraded 2015-06-10 13:10:32 and cairo 2015-06-10 13:10:44 hmm 2015-06-10 13:11:26 freetype is now 2.6 + Infinality patches 2015-06-10 13:12:33 i dont have X here to try now :( 2015-06-10 13:13:37 bah no hurry ;) 2015-06-10 13:22:53 clandmeter: i'll leave that up to you, i can imagine it may come in handy for people to know the hostnames needed to address individual PoP's too 2015-06-10 13:25:55 ncopa: i use weechat on alpine but remotely under tmux and w/ rxvt-unicode on ArchLinux... i dont have that rendering problem 2015-06-10 13:33:52 clandmeter: btw we also offer rsync and FTP 2015-06-10 13:34:56 and ipv6 on the NL, DE and US east coast mirror 2015-06-10 13:36:00 hrak: i added the rsync links in our mirror file 2015-06-10 13:57:58 AmatCoder: its fontconfig 2015-06-10 14:25:01 new to alpine, trying to get docker to work. when i pull an image i get this upon building, stderr: chmod /bin/mount: permission denied...its being ran as root and my user is root any ideas? 2015-06-10 14:27:06 hello 2015-06-10 14:27:54 have someone an idea with RPi 1 or 2 to have full acces of memory managment ? 2015-06-10 14:28:19 on RPi 2 i only have 770Mo of Ram with alpine 2015-06-10 14:28:34 if i install raspbian, i have 1Go 2015-06-10 14:30:20 prolly the gpu eating memory (like on rpi 1) 2015-06-10 14:31:38 found this grsec: denied chmod +s, not sure how to disable this yet 2015-06-10 14:31:47 Diftraku i change gpu_mem=16 2015-06-10 14:32:06 the result is the same 2015-06-10 14:35:27 Chipster1: its tmpfs install i guess? 2015-06-10 14:35:45 which means some memory is reserved for / 2015-06-10 14:35:55 i know 2015-06-10 14:36:06 rasbian is a disk install 2015-06-10 14:36:11 i use free -m 2015-06-10 14:36:23 total used free shared buffers 2015-06-10 14:36:23 Mem: 745 35 710 8 0 2015-06-10 14:36:57 with raspbian i have ~1Go 2015-06-10 14:37:41 a bug with \t :p 2015-06-10 14:38:10 what does grep MemTotal /proc/meminfo say? 2015-06-10 14:39:02 wait 2015-06-10 14:39:09 i try another thing 2015-06-10 14:41:33 clandmeter MemTotal: 763792 kB 2015-06-10 14:43:46 fabled: any idea? 2015-06-10 14:47:21 as I read it is a kernel bug 2015-06-10 14:55:54 can Xorg hotplug input devices for a system using mdev, instead of udevd? 2015-06-10 14:59:57 ACTION thinks no 2015-06-10 16:49:07 for some strange reason, IPv4+IPv6 dualstack on the same interface is not working for me 2015-06-10 16:49:18 is this a known problem? 2015-06-10 16:51:06 did you modprobe ipv6 ? 2015-06-10 16:51:50 myhost:~# lsmod | grep ipv6 2015-06-10 16:51:50 ipv6 331675 27 bridge,[permanent] 2015-06-10 16:51:55 so it is loaded, yes 2015-06-10 16:52:47 I've put a static configuration for both IPv4 and IPv6 on the same interface (eth0) into /etc/network/interfaces 2015-06-10 16:53:07 IPv4 works flawlessly, but for IPv6, I only get a link-local address o_O 2015-06-10 16:55:39 neither the IPv6 address nor the IPv6 gateway/default route is set 2015-06-10 16:56:43 i used openvswitch with ip4 & ip6 2015-06-10 16:59:45 well that doesn't really help me :/ 2015-06-10 16:59:48 my interfaces are like http://hastebin.com/epanuyitej.hs 2015-06-10 17:00:26 I set 2 x ipv6 addresses & 1 sometimes 2 ip4 onthe same interface 2015-06-10 17:01:57 actually 2 of each at the moment - i missed off the 2nd ip4 by mistake 2015-06-10 17:04:40 back 2015-06-10 17:05:10 BitL0G1c: I' using the config from here: http://wiki.alpinelinux.org/wiki/Configure_Networking#Example:_Dual-Stack_Configuration 2015-06-10 17:05:29 btw, it seems to work when I manually assign an IPv6 address using ip -6 add 2015-06-10 17:06:04 but of course, it should also work without manual intervention, and even without special if-up action 2015-06-10 17:07:21 if you are using bridge-utils - you also need to 'apk add bridge' & ' rc-update add bridge' 2015-06-10 17:09:02 & rc-service bridge start 2015-06-10 17:19:19 btw, how can I list the files and directories that a package contains? 2015-06-10 17:33:02 http://pkgs.alpinelinux.org/contents 2015-06-10 17:35:40 buerofalke: or tar -tzf some.apk 2015-06-10 17:36:44 ok, thx 2015-06-10 17:37:26 sad you don't use pacman :/ 2015-06-10 17:42:40 and how can I find out what package a file belongs to? 2015-06-10 17:52:42 nevermind, found it 2015-06-10 17:52:57 apk info --who-owns /usr/sbin/brctl 2015-06-10 17:54:31 oh, and to show the content of a package: 2015-06-10 17:54:31 apk info -L bridge 2015-06-10 20:53:13 ahills: i noticed you're building ghc for alpine too (saw you in #ghc), i have all my stuff in dockerfiles for stage2 generation etc... and am just working on compiling natively off of the stage2 build to get to an apk, currently working on that atm, have dynamic linking working as well, or seemingly working 2015-06-10 20:56:29 been working on this on/off for the past week, not a ton though 2015-06-11 00:11:21 thank you for the welcome, folks 2015-06-11 00:47:16 Is there any sensible 2-factor auth method for openssh on alpine? I wanted to use oath-toolkit and was already done building the package when i realised that openssh is built without pam support 2015-06-11 02:25:07 tze: I've not gotten around to releasing oath-toolkit as a package 2015-06-11 02:25:23 tze: I've also started on porting a lot of the yubico software to Alpine 2015-06-11 02:25:54 and a variety of other pam modules 2015-06-11 02:26:29 tze: I think having a openssh build with PAM support is reasonable but I would want to check wiht ncopa and others on what they think 2015-06-11 02:26:42 That's great i didn't even realize that. I actually build an APKBUILD for oath-toolkit. 2015-06-11 02:27:01 awesome 2015-06-11 02:27:22 That would be great, i didn't want to bother building my own openssh package with pam, that would just make everything unclean in my book 2015-06-11 02:27:47 if you want me to help / test / document anything, please let me know 2015-06-11 02:28:59 systemkor: Did you get around to build a package for xmlsec? It is needed so oath-toolkit is able to be built with pkcs support. It is acutally included with libreoffice-common, but that installs so much other dependencies that it's kind of ugly to use it in this context 2015-06-11 02:29:16 Although i personally don't need pkcs support :) 2015-06-11 02:30:19 I don't remember, that said if it is simple enough to write new code to overcome erroneous dependencies I might be down to do that 2015-06-11 02:31:41 Alright, i'll look into it in the next couple of days and let you know. Although it would make sense to clarify if me make an openssh package with pam support. Otherwise the work is kind of pointless 2015-06-11 02:35:31 "to clarify if me make an openssh"? 2015-06-11 02:35:49 sorry not quite understanding what you mean 2015-06-11 02:36:14 are you saying I (i.e. sysmtmkor) should build OpenSSH with PAM support? 2015-06-11 02:38:34 if so I'll gladly try my hand at it 2015-06-11 03:49:52 ncopa: with isl at 0.14.x (why is it 0.14.1 and not 0.14.2 btw?) can i update the llvm/clang build with support for polly? it actually doesn't require any porting work for musl and works just fine. http://polly.llvm.org/ 2015-06-11 03:50:51 just noticed i asked that in alpine-linux and not alpine-devel. whoops. still. 2015-06-11 03:52:28 ncopa: i've discovered i also need to fix update-kernel and update-extlinux to work with EFI, but i'm potentially too lazy to do it. there's a lot of hard coded stuff going on. funny enough, your syslinux package already installs the EFI stuff, the other scripts just don't handle that kind of setup 2015-06-11 05:46:03 mitchty: unfortunately all my projects are paused, I'm travelling to Sweden for midsummer 2015-06-11 05:57:50 morning 2015-06-11 05:58:17 ttilley: i didnt find any isl-0.14.2. http://isl.gforge.inria.fr/ 2015-06-11 06:15:15 ncopa: odd. maybe it's tagged in source control but not released as a tarball? there are packages for it in ubuntu either way. 2015-06-11 06:17:48 ncopa: also, good morning :) 2015-06-11 06:18:01 2am here 2015-06-11 06:19:37 still coming off a WWDC high. they're open sourcing swift 2.0, including the swift standard library. that's pretty exciting stuff for a NeXTStep fanboy like myself. i'm going to have fun writing QT5 binding for it on linux 2015-06-11 06:20:42 ...i hope it doesn't depend on too many GNUisms and is easy to port to musl 2015-06-11 06:21:28 it has a BSD based OS as its primary platform so i hope not, but the same can be said about LLVM... -_- 2015-06-11 06:33:43 i saw about swift yes 2015-06-11 06:33:56 and yes it is interesting 2015-06-11 06:39:52 i'm SUPER FUCKING EXCITED 2015-06-11 06:39:54 =D 2015-06-11 06:40:18 ACTION does his excited dance... which actually involved zero movement, staring at an IRC window by the computer 2015-06-11 06:43:01 ncopa: anyways. should i add support for polly to the LLVM/Clang build in edge? it has no effect for those who don't use it, but statically compiled into the utilities the options for loop optimization are there without loading polly as a plugin. useful stuff. 2015-06-11 06:44:00 it might make upgrading isl tricky in the future, is the only issue. i don't know if gcc and clang will always want the same version like they do currently 2015-06-11 07:14:52 ttilley: i dont have any major objection against polly 2015-06-11 07:15:03 gcc has a patch that makes isl optional 2015-06-11 07:15:09 runtime optional 2015-06-11 07:15:13 oh? neat 2015-06-11 07:15:17 it will dlopen it 2015-06-11 07:15:17 i didn't know that 2015-06-11 07:15:26 that makes it easier to upgrade isl 2015-06-11 07:15:41 it seems pretty essential for graphite so i didn't expect that kind of flexibility 2015-06-11 07:15:43 you upgrade isl, at which point gcc would break without this patch 2015-06-11 07:15:51 then you rbuild gcc 2015-06-11 07:16:14 nice 2015-06-11 07:16:45 http://git.alpinelinux.org/cgit/aports/tree/main/gcc/074_all_gcc5_isl-dl.patch 2015-06-11 07:16:45 ...clang, however, will straight up break if there's API incompatibility in ISL between versions and expected symbols are not present 2015-06-11 07:17:15 if polly is enabled statically that is. and it should be, otherwise using polly is insanely complicated. 2015-06-11 07:17:31 you have to make sure clang loads it as a plugin if it's not statically in clang itself 2015-06-11 07:17:40 and that's just too much effort and complexity 2015-06-11 07:17:48 IMHO 2015-06-11 07:17:56 so what does polly give us? 2015-06-11 07:18:14 the same kind of functionality as graphite in gcc. 2015-06-11 07:18:39 polyhedral loop optimizaitons 2015-06-11 07:19:10 ok 2015-06-11 07:19:31 re clang and default hardened 2015-06-11 07:19:55 you have patches for default ssp and default pie 2015-06-11 07:20:19 i have patches for default PIE, but not SSP. i don't fully understand the specifics of how SSP works with musl yet 2015-06-11 07:20:21 but i think we might also need link options for bind now and relro by default 2015-06-11 07:20:35 that, i can do. and is easy. 2015-06-11 07:20:49 the binutils related code is very nice to work with in LLVM 2015-06-11 07:21:02 i can have it set those as defaults if running on alpine 2015-06-11 07:21:19 (which is nice for me, specifically, because i don't want to break any existing tests in the suite) 2015-06-11 07:22:28 ncopa: those LTO tests that fail everywhere i've run them and so i disable them in the build... btw... in 3.7 they pass just fine on musl. very weird. i dont know why they fail in 3.6. 2015-06-11 07:23:05 morning 2015-06-11 07:23:36 maybe it is broke in 3.6 :) 2015-06-11 07:23:58 ssp is basically only -fstack-protector enabled by default 2015-06-11 07:24:37 ncopa: the libc++ test suite doesn't even run on 3.6 due to the python test runner being legitamitely broken. they accidentally backported syntax from 3.7 that doesn't work on 3.6. whoops. so yeah, i wouldn't be amazed if it was something like that. 2015-06-11 07:25:15 though i'm afraid that NOBODY has actually EVER run the test suite for libc++ on 3.6 2015-06-11 07:25:25 :) 2015-06-11 07:27:01 ncopa: still, my OCD prevents me from being comfortable installing something that doesn't pass its test suite. i'm sure you understand. just disabling the LTO tests stressed me out a bit. 2015-06-11 07:27:20 yeah 2015-06-11 07:27:22 and i only disabled them because they fail EVERYWHERE. alpine, gentoo, ubuntu... 2015-06-11 07:27:31 patched or not 2015-06-11 07:27:43 sounds like the test itself is broke 2015-06-11 07:28:25 ScrumpyJack: morning 2015-06-11 07:28:41 hi ScrumpyJack 2015-06-11 07:28:50 oh... reminds me 2015-06-11 07:29:02 ScrumpyJack: didnt you have a bunch of patches send to alpine-devel 2015-06-11 07:29:17 ncopa: my understanding is that with SSP on musl, it involved linking in a shared version of the SSP code that wouldn't otherwise be a concern. hence why i didn't want to touch it yet without fully understanding what was going on. 2015-06-11 07:30:50 i suppose its just to test with -fstack-protector manually 2015-06-11 07:32:23 i mean, making -Wl,-z,now -Wl,-z,relro a default is simple. but if the way SSP works is fundamentally different on musl based systems, i want to understand that first... and currently i don't 2015-06-11 07:32:28 ^^; 2015-06-11 07:33:02 i dont think it is fundamentally different 2015-06-11 07:35:17 yes. ssp works same everywhere 2015-06-11 07:35:21 i want to catch a WWDC session on swift performance in about 8 hours so i should probably sleep... but expect a bug filed in the near future implementing your suggestions about -Wl,-z,now -Wl,-z,relro as well as enabling polly in clang (the llvm equivalent of gcc's graphite) 2015-06-11 07:35:54 no offense, but if i have to sign up for a mailing list to submit patches i'm probably going to be too lazy to do it. i'll use the bug system. :p 2015-06-11 07:40:03 fabled: i'm excited you got a version of gcc using isl 0.14 in edge btw. thanks for that :) 2015-06-11 07:45:50 ncopa: yes, i did. silly me sent them to the wrong list. 2015-06-11 08:06:53 ttilley: i feel with you re sign up to ml to send patches... 2015-06-11 08:07:09 sent 100+ patches to different upstreams 2015-06-11 08:07:29 i don't think you need sign-up for alpine-aports 2015-06-11 08:08:12 ncopa: one project i wanted to send a musl patch to required subscription to a ML to do so. instead, i sent it to gentoo and the gentoo dev handled the upstream bullshit for me. win/win situation if you ask me. anyways, i SHOULD be asleep. heh 2015-06-11 08:08:55 ttilley: good night! 2015-06-11 08:22:20 Woo, a channel. 2015-06-11 08:23:47 hi JamesTK 2015-06-11 08:23:55 hithere 2015-06-11 11:27:17 hm 2015-06-11 11:27:31 I get this error compiling: 2015-06-11 11:27:48 /usr/include/sched.h:75:17: error: expected identifier before string constant void free(void *); 2015-06-11 11:28:23 weird because it works a few weeks ago... 2015-06-11 11:28:47 gcc 5.1 issue? 2015-06-11 11:35:16 hi guys, got this error, but do not know what exactly goes wrong. "hostname:~# apk upgrade 1 errors; 13 MiB in 27 packages" 2015-06-11 11:35:38 anyway to get what the error is? 2015-06-11 11:57:52 xen_roger: did you try apk fix? 2015-06-11 11:58:22 you fixed it already 2015-06-11 12:05:35 clandmeter: sorry 4 late response... yeah fix did not work 2015-06-11 12:06:11 clandmeter: ncopa gives the hint to remove all and reinstall because of renaming freeradius pkg 2015-06-11 12:06:22 wich did the job for me 2015-06-11 12:27:11 ncopa, what did you do for font revolution in console ? I would try same to see if it improves newsbeuter ! 2015-06-11 12:32:19 vkrishn: i did apk upgrade 2015-06-11 12:32:33 but clandeter reverted the revolution 2015-06-11 12:39:54 hmm, ok 2015-06-11 12:41:44 i did? 2015-06-11 12:41:59 i only reverted fontconfig 2015-06-11 12:42:25 freetype should still be updated including patches 2015-06-11 12:45:10 does it improve console fonts ? 2015-06-11 12:45:48 i think so 2015-06-11 12:46:09 although i think previous infinality was looking better. 2015-06-11 12:46:28 how to add, guide pls 2015-06-11 12:46:34 apk add freetype ? 2015-06-11 12:47:56 apk add freetype@edge/main 2015-06-11 12:49:48 that does not pull "freetype-infinality" 2015-06-11 12:50:08 apk search freetype, show its available 2015-06-11 12:50:23 its merged into freetype now 2015-06-11 12:50:30 but only in edge ofc 2015-06-11 12:50:48 so if you are on edge, a apk upgrade should do 2015-06-11 12:51:09 vkrishn-edge-x86_64:/dev/shm# apk add freetype 2015-06-11 12:51:10 (1/2) Installing libpng (1.6.17-r0) 2015-06-11 12:51:10 (2/2) Installing freetype (2.6-r0) 2015-06-11 12:51:25 correct 2015-06-11 12:51:27 and restart X 2015-06-11 12:51:33 oooohhh 2015-06-11 12:52:06 I thought it would improve in console... gee sorry 2015-06-11 12:52:21 ? 2015-06-11 12:52:38 you dont have X running on alpine? 2015-06-11 12:52:59 no 2015-06-11 12:53:04 :) 2015-06-11 12:53:19 nor do I plan, except for testing maybe 2015-06-11 12:53:43 atleast for next 2 AL cycle 2015-06-11 12:53:43 what are you using as de? 2015-06-11 12:53:58 and dist? 2015-06-11 12:54:56 logged into clandmeter's AL box via ssh, using knoppix de 2015-06-11 12:55:05 ;) 2015-06-11 12:55:36 you need to fix font rendering within knoppix 2015-06-11 12:57:59 http://i.imgur.com/tHeGCxI.png, thinking the wierd unaligned columns gets fixed 2015-06-11 12:59:23 vkrishn: thats probably term related 2015-06-11 13:00:04 there was some discussion about it yesterday. you can read backlog 2015-06-11 13:00:36 ok 2015-06-11 13:00:54 did you try running it in tmux? 2015-06-11 13:01:20 same :( 2015-06-11 13:11:46 great, 'shed.h' issue was fixed with this patch: 2015-06-11 13:11:49 https://raw.githubusercontent.com/voidlinux/void-packages/c3d0cf462b9e01896ead8a5a839c065ca3e5bfdb/srcpkgs/mame/patches/musl-sched.patch 2015-06-11 15:27:05 ahills: no big deal, i'm just cleaning up my dockerfiles and working on getting to apk's for ghc/cabal/happy/alex 2015-06-11 16:07:38 systmkor: Oh no, not necessarily i mean we should discuss in general with ncopa if pam support for ssh is even wanted in general 2015-06-11 16:58:38 tze: can it be built as a module? 2015-06-11 17:07:59 ahills: That's a good question, i'm not to familiar with the openssh build options, but i think probably not. One Option would be to just have a seperate openssh-pam package. Although that's not really ideal as both have to be maintained. 2015-06-11 18:37:50 hello all 2015-06-11 18:38:58 i'm trying to use sed command and apparently it has some serious differences with gnu sed. is there any documentation outlining those differences? what can we use and what we can't and what needs to be kept in mind? 2015-06-11 18:47:29 you can install gnu sed 2015-06-11 18:47:57 yes, but if you want to write portable scripts it's more valuable to learn what's not portable 2015-06-11 18:50:46 Busybox sed does not support +7d for example 2015-06-11 18:59:58 the only other issue I had was in ~/aports/testing/bitcoin/bitcoin.post-install - is tricky to do substitution if the item to replace includes an = 2015-06-11 19:00:45 the +7d with in ~/aports/testing/psad/APKBUILD 2015-06-11 19:05:52 so i'm not 100% sure where to ask this, but i've managed to build ghc 7.10.1 for alpine linux and it seems to work, but how should I go about getting this into alpine linux? ghc had to be bootstrapped from an ubuntu image and then built inside alpine linux so building from source has a bit of a cicular dependency 2015-06-11 19:06:40 i've done it all in dockerfiles though so its pretty straightforward to repeat, just takes ages 2015-06-11 20:43:55 BitL0G1c: i did, but i'd rather not if there's a way to accomplish the same using the standard tools 2015-06-11 21:14:31 ianbytchek - I've not needed to use gnu awk or sed recently 2015-06-11 22:33:57 does Alpine use LibreSSL or OpenSSL by default? 2015-06-11 22:43:32 flatr0ze: OpenSSL 2015-06-11 22:44:09 flatr0ze: I need to update LibreSSL & try full build with it and see how well it holds up, sorry been swamped with work and then got sick recently 2015-06-11 22:59:44 ianbytchek: busybox sed is supposedly POSIX compliant, so the POSIX sed spec should have what you're looking for 2015-06-11 23:03:17 all good systmkor, thanks for the answer 2015-06-11 23:29:59 question with abuild, is this telling me that i'm missing a library in my apkbuild? 2015-06-11 23:30:03 https://gist.githubusercontent.com/mitchty/4dfd1657cfcef04a67ec/raw/98e7f69b7afd7beff016bd8886e8bed07d387939/gistfile1.txt 2015-06-11 23:35:34 or dependency rather i suppose 2015-06-12 00:59:38 systmkor: i think libressl is fundamentally the wrong approach 2015-06-12 01:00:49 with libressl, we are effectively polishing a turd. it may be better to build something that is compatible with openssl, using proven cryptographic blocks like nettle 2015-06-12 01:01:16 i do not think that using libressl over openssl will improve security in any way 2015-06-12 01:01:22 nettle? seriously? 2015-06-12 01:01:31 how about tomcrypt 2015-06-12 01:01:44 well, that could work too :) 2015-06-12 01:01:59 i am just saying libressl has introduced their own bugs 2015-06-12 01:02:48 kaniini, maybe you've missed the discussions on #musl 2015-06-12 01:02:58 kaniini: ehhhh what you are saying about LibreSSLL doesn't really comply with reality 2015-06-12 01:03:03 *LibreSSL 2015-06-12 01:03:11 systmkor: it will never be included in alpine just give up 2015-06-12 01:03:29 unless openssl does something really really bad 2015-06-12 01:03:35 like heartbleed times 1000000000 2015-06-12 01:03:37 our aim for tls is to design a dead-simple, hard-to-misuse (stupid usage is actually harder to program than correct usage) tls library-in-a-command 2015-06-12 01:03:44 there is no justification for it 2015-06-12 01:04:17 there are several planned implementation phases 2015-06-12 01:04:38 dalias: this seems reasonable, but how will compatibility with current cryptographic stacks be achieved 2015-06-12 01:05:08 first phase is defining the API as an external command that runs in its own process, and which you can invoke yourself or use a simple library to invoke for you 2015-06-12 01:05:26 in this phase, the internal implementation can use whatever legacy TLS lib you want 2015-06-12 01:05:27 and more importantly, i think if you're going to do that -- it is a reasonable idea to provide an API that allows you to sandbox other components as well easily 2015-06-12 01:05:50 well the problem with that is, openssl is used for way more than tls ;) 2015-06-12 01:05:50 kaniini: http://i.imgur.com/Hej4ZKh.gifv 2015-06-12 01:06:24 systmkor: yes, i will deal with it, in much the way i will deal with any inclusion of systemd :) 2015-06-12 01:06:30 then once it's working, the next phase is a clean reimplementation of the internals 2015-06-12 01:06:59 kaniini: okay, so your just going to throw a childish temper tantrum? 2015-06-12 01:07:12 systmkor: no, it will just not make it past testing 2015-06-12 01:07:21 kool story bro 2015-06-12 01:07:22 systmkor: unless real documentation is provided proving it is superior 2015-06-12 01:07:36 once there's a clean implementation, the possible third-phase is making it possible to link into your program and run the TLS in one or more threads rather than as an external process per connection 2015-06-12 01:07:57 that trades some safety for some resources 2015-06-12 01:08:04 systmkor: yes, "kool story" indeed. hint: if we wanted to use libressl, it would be done by now because we already had to face an ABI bump with openssl 1.0.2. 2015-06-12 01:08:21 yah, and? 2015-06-12 01:08:31 systmkor: our decision was to stick with openssl. 2015-06-12 01:08:32 kaniini, is there a reason you're so opposed to switching to libressl? 2015-06-12 01:08:41 kaniini: and? 2015-06-12 01:08:56 dalias: yes, i am not convinced it is [in the real world] an improvement over openssl security-wise 2015-06-12 01:08:57 i don't like libressl, but it seems a lot less-bad then openssl 2015-06-12 01:09:13 kaniini, it's a clear improvement in removing UB and nastiness 2015-06-12 01:09:15 dalias: how much of that is actual matter and how much of it is fluff though 2015-06-12 01:09:29 sure, but openssl is getting similar improvements too 2015-06-12 01:09:46 and openssl has far better funding (linux foundation core initiative) than libressl 2015-06-12 01:09:49 have they completely rm'd all code related to export ciphers? 2015-06-12 01:10:14 consider what you are saying, though 2015-06-12 01:10:25 rm'ing massive piles of stuff is a knee-jerk reaction 2015-06-12 01:10:30 no it's not 2015-06-12 01:10:34 kaniini: despite the current funding which I argue is a point for OpenSSL 2015-06-12 01:10:47 OpenBSD has had a much better track record for projects longterm than OpenSSL 2015-06-12 01:10:49 removing dangerous legacy code that has no modern use is good practice 2015-06-12 01:10:59 dalias: +1 that 2015-06-12 01:11:02 dalias: that is a valid point 2015-06-12 01:11:04 it's not a reaction, it's a proper precaution 2015-06-12 01:11:16 same principle as musl here 2015-06-12 01:11:20 but libressl introduced bugs by aggressively removing code 2015-06-12 01:11:25 kaniini: such as OpenSSL support https://en.wikipedia.org/wiki/EBCDIC 2015-06-12 01:11:35 since you know everyone still uses that all the time 2015-06-12 01:11:36 systmkor: who fucking cares if openssl supports EBCDIC 2015-06-12 01:11:43 systmkor: alpine does not support EBCDIC 2015-06-12 01:11:50 systmkor: thusly openssl on alpine is not built with said support 2015-06-12 01:12:06 if there is crypto code whose ONLY POSSIBLE PURPOSE is to make you vulnerable, it needs to be removed 2015-06-12 01:12:07 that's not quite how it works but sure 2015-06-12 01:12:07 systmkor: same for OPENSSL_VMS 2015-06-12 01:12:16 systmkor: this is alpine linux, not alpine vms 2015-06-12 01:12:42 at any rate, any major change like that is going to require a serious burden of proof 2015-06-12 01:12:51 "oh, all the packages build with it" isn't going to be enough, sorry 2015-06-12 01:13:06 *nod* 2015-06-12 01:13:27 well yah 2015-06-12 01:13:27 i'm also unconvinced by your CII funding argument tho 2015-06-12 01:13:29 that's a given 2015-06-12 01:13:44 dalias: sure 2015-06-12 01:13:45 all along openssl has been about getting bureaucratic funding for doing bullshit 2015-06-12 01:13:54 that's what got us into this mess 2015-06-12 01:14:00 I just wanted to get LibreSSL working so we could test it, if nothing comes of it then w/e 2015-06-12 01:14:00 they were a FIPS consulting company 2015-06-12 01:14:02 dalias: my larger argument is that openssl is a turd and replacing it with a polished turd is well, it's still a turd 2015-06-12 01:14:42 i think a better study would be to see how many packages can be built with alternatives to openssl (NSS is a good target to port to) 2015-06-12 01:15:00 kaniini, i see it more as a turd with tapeworms in it, and rather than polishing, removing the tapeworms 2015-06-12 01:15:03 it's still pretty nasty 2015-06-12 01:15:11 but not a major health hazard ;) 2015-06-12 01:15:12 (why NSS? it's API is very similar to OpenSSL, because OpenSSL is a clone of NSS to begin with) 2015-06-12 01:15:36 dalias: and my point is they haven't removed all tapeworms because we cannot know that 2015-06-12 01:15:40 *nod* 2015-06-12 01:15:58 either way you're getting toxiplasmosis 2015-06-12 01:16:04 so the real fix is to get rid of openssl altogether 2015-06-12 01:16:07 yes 2015-06-12 01:16:11 hince why we decided to not jump on the libressl train 2015-06-12 01:16:13 and NSS is not a good replacement 2015-06-12 01:16:23 it is not, but it may be a good stepping stone 2015-06-12 01:16:27 due to similar API 2015-06-12 01:16:42 our program (#musl) is to get rid of the need for an api 2015-06-12 01:16:45 for TLS, that is 2015-06-12 01:17:05 the problem is supporting pre-existing ecosystem 2015-06-12 01:17:11 i like the idea of having a TLS helper 2015-06-12 01:17:22 you simply trade in your original socket for a new socket 2015-06-12 01:17:33 and use it like you would an unencrypted socket 2015-06-12 01:17:54 right, that is how we implemented TLS in charybdis (the ircd freenode uses) 2015-06-12 01:18:05 and the code running on the other end of your new socket does all the crypto work 2015-06-12 01:18:08 the rationale for that was to allow TLS processing to not block the application code 2015-06-12 01:18:15 which is nice 2015-06-12 01:18:20 _including_ cert checking 2015-06-12 01:18:27 yep 2015-06-12 01:18:31 if you want to override cert checking or do anything fancy with it... 2015-06-12 01:18:33 and then you have a control socket 2015-06-12 01:18:35 you have to use a nastier api :) 2015-06-12 01:18:37 which allows you to get all of that 2015-06-12 01:18:41 yep 2015-06-12 01:18:47 there's already code which does this if you want to look 2015-06-12 01:18:52 where is it? 2015-06-12 01:19:01 https://github.com/atheme/charybdis/blob/master/ssld/ssld.c 2015-06-12 01:19:04 this sounds almost exactly like the design we've been working on 2015-06-12 01:19:14 https://github.com/atheme/charybdis/blob/master/src/sslproc.c 2015-06-12 01:19:25 it's tied to a piece of junk called libratbox 2015-06-12 01:19:31 but, it is probably a good starting point 2015-06-12 01:19:35 since it works in the real world 2015-06-12 01:19:53 :-p 2015-06-12 01:19:56 a cool feature is 2015-06-12 01:20:01 if some wanker tries to do an openssl exploit 2015-06-12 01:20:01 yeah it might be a good place to see how it was done before 2015-06-12 01:20:10 the ircd does not die 2015-06-12 01:20:14 and if there are good ideas to carry away, or mistakes to avoid repeating 2015-06-12 01:20:14 it just loses some SSL clients 2015-06-12 01:20:33 with our design, it's just one client -- the one who exploited it 2015-06-12 01:20:42 every tls connection is its own process 2015-06-12 01:21:01 that could be problematic 2015-06-12 01:21:05 compared to kernel socket buffers, the resource usage of a process is tiny 2015-06-12 01:21:08 because PID is only 16-bit namespace on linux 2015-06-12 01:21:14 it's 29 bit 2015-06-12 01:21:17 oh 2015-06-12 01:21:18 well 2015-06-12 01:21:22 still 2015-06-12 01:21:22 the default max is just 32768 2015-06-12 01:21:39 but yes 2015-06-12 01:21:43 its probably better to just have a pool of ssl workers 2015-06-12 01:22:04 or tls workers rather 2015-06-12 01:22:20 our actual motive for doing it that way 2015-06-12 01:22:28 was to try to sidestep a GPL problem ;) 2015-06-12 01:22:40 for the vast majority of things that aren't bound by connections, it really makes more sense just to do a process per conn 2015-06-12 01:22:52 only servers with lots of clients have to worry about being more efficient 2015-06-12 01:23:25 yes, you could have both modes 2015-06-12 01:23:32 the eventual thread-based option will definitely do pooling 2015-06-12 01:23:45 ehhhh 2015-06-12 01:23:46 rather it will do N threads where N i wouldnt use threads ;) 2015-06-12 01:23:56 i would just use processes. 2015-06-12 01:23:58 keep it simple 2015-06-12 01:24:03 threads are much simpler 2015-06-12 01:24:14 i dont know if simplicity is needed because, 2015-06-12 01:24:20 it's just socketpair(2) man 2015-06-12 01:24:25 well you're the one who said "keep it simple" :) 2015-06-12 01:24:47 well what i mean is processes provide better resiliency 2015-06-12 01:25:19 yes but most of the value of doing processes disappears if you've got multiple users' connections going thru one process 2015-06-12 01:25:19 systmkor: however, my concern with libressl is two-fold. 2015-06-12 01:25:24 yes 2015-06-12 01:25:30 there's a risk of exploits compromising someone else's data 2015-06-12 01:25:32 systmkor: 1. i do not want FooSSL to be a "long-term project", i want it to die 2015-06-12 01:25:48 systmkor: 2. we do not yet know which has less tapeworms 2015-06-12 01:26:15 for 1. can you unpack or re-word 2015-06-12 01:26:38 systmkor: for (1), my point is that reducing the size of the entire OpenSSL ecosystem is better 2015-06-12 01:26:55 systmkor: OpenSSL in it's entirety should die, and by die, i do not mean replaced by LibreSSL 2015-06-12 01:27:02 as a strategy I agree 2015-06-12 01:27:07 i agree libressl is a bandaid 2015-06-12 01:27:11 but possibly a very useful 2015-06-12 01:27:13 one 2015-06-12 01:27:14 it may be for the short-term that libressl is a better bandaid 2015-06-12 01:27:21 we still need to determine this 2015-06-12 01:27:46 at the time we bumped to 1.0.2 openssl looked more viable because libressl had just had a bunch of problems in the linux port and it broke API compatibility 2015-06-12 01:28:14 I haven't said we should ditch OpenSSL and do LibreSSL full stop, I just want it working for people to test with that's really it, if it ends up being safer then move there if it ends up being useless I'll ditch it 2015-06-12 01:28:18 i don't think the linux port was meant to be ready-for-deployment yet at that time, was it? 2015-06-12 01:28:31 no it was the official one 2015-06-12 01:28:44 systmkor: i would prefer to at least keep openssl as it is right now until apk-tools' use of libcrypto is replaced by something else like libtomcrypt 2015-06-12 01:29:04 getting rid of libcrypto use is a good goal 2015-06-12 01:29:21 yah sure, I never said anything about apk-tools with respect to OpenSSL or LibreSSL 2015-06-12 01:29:26 using an ssl/tls lib for crypto that doesn't need all the tls protocol complexity is rather bad 2015-06-12 01:29:34 systmkor: yes, but my point with this is 2015-06-12 01:29:39 the worst part about all these crypto libs.... 2015-06-12 01:29:51 they do runtime-initialization of big tables 2015-06-12 01:30:04 our core system software (apk-tools, abuild, etc.) is dependent on openssl libcrypto 2015-06-12 01:30:11 this makes them slow and bloated for putting TLS in its own process 2015-06-12 01:30:21 libressl libcrypto may have changes that wind up bricking apk-tools somehow 2015-06-12 01:30:26 if they correctly used static const [] tables 2015-06-12 01:30:35 all that would be shareable and load instantly 2015-06-12 01:31:31 dalias: why not use cryptlib as backend? 2015-06-12 01:31:32 the tables also have all sorts of thread-safety bugs 2015-06-12 01:31:33 kaniini: yah and? I didn't say ncopa or others should have to maintain it I just said I wanted to try to get it to work as a whole system to see if there were issues and address them if need to be 2015-06-12 01:31:41 kaniini, i'm not really familiar with it 2015-06-12 01:31:57 does it get this right? 2015-06-12 01:32:08 a proper crypto lib should have 0 bytes of data/bss 2015-06-12 01:32:30 dalias: i am not 100% sure, but i believe it builds tables on demand instead of putting it in .data/.bss 2015-06-12 01:32:46 *sigh* 2015-06-12 01:32:48 so idiotic 2015-06-12 01:33:00 build-on-demand is exactly the wrong thing 2015-06-12 01:33:05 the tables belong in .text 2015-06-12 01:33:19 where they take no time to build and use no memory 2015-06-12 01:33:26 hmm, true 2015-06-12 01:33:47 building at startup makes the tls helper processes start slow and use lots of memory 2015-06-12 01:34:01 dalias: by puting it in .text are you just moving the memory that was taken up by .data and moving it to the .text pages? 2015-06-12 01:34:25 systmkor, you're sharing it with the fs cache and with all processes using the library 2015-06-12 01:34:37 dalias: ahhhhhhhhhhh 2015-06-12 01:34:46 dalias: it may be possible that it statically has the tables it needs 2015-06-12 01:34:50 so there's only one copy of the tables rather than hundreds of copies 2015-06-12 01:34:53 i am not sure, it is an embedded-oriented library 2015-06-12 01:35:16 it is commonly used with newlib on bare metal 2015-06-12 01:35:27 so i would figure it builds on demand due to memory requirements 2015-06-12 01:35:36 kaniini, embedded-oriented is less likely to do this right because they often don't have mmu/memory-sharing 2015-06-12 01:35:46 Hello all. I have a docker file which basically gets wget, downloads the official Oracle JRE (1.8.45 b14) and puts it into /usr/local/jdk1.8.0_45 then just symlinks to /usr/local/jdk. I add the path (which exists) /usr/local/jdk/bin to the path and in /usr/local/jdk/bin there is the java binary with 755 permissions. 2015-06-12 01:37:02 Yet... java shows "not found" when I try to run it. If I jump into an interactive terminal with /bash/sh I can cd to /usr/local/jdk/bin or /usr/local/jdk1.8.0_45/bin and SEE java... yet when I try to start it it says "not found". Any thoughts on where I have gone wrong? It is quite obviously there. 2015-06-12 01:41:49 kaniini: so would how do you think we should try to measure, test, verify TLS/crypto libs for Alpine? 2015-06-12 01:42:05 *so how do you think 2015-06-12 01:42:43 trco, i would assume your java binary is linked to glibc which is not-found :) 2015-06-12 01:43:02 hmm 2015-06-12 01:43:12 though you can overcome this, I think there is details on the wiki 2015-06-12 01:43:13 systmkor, a lot of work on this has already been done in #musl (comparing TLS libs to see how bad they are) 2015-06-12 01:43:16 nothing good was found 2015-06-12 01:43:37 dalias: oh, well thank you :P 2015-06-12 01:44:18 dalias: do you have a link of that data? 2015-06-12 01:44:44 dalias: ... I see! Sometimes you just need a little bump in the right direction 2015-06-12 01:44:58 trco: good luck! :D 2015-06-12 01:45:02 systmkor, not right off, but i think graphitemaster has it 2015-06-12 01:45:42 m'kay 2015-06-12 01:46:31 dalias: you have happened to work wiht aufs, union-fs, or overlay-fs? 2015-06-12 01:47:18 no, i've noted that none of them get file identity (inode) semantics right tho :( 2015-06-12 01:47:48 how so? 2015-06-12 01:47:50 opening and editing a file on the base fs always looks (to applications) like deleting it and replacing it with a new file, rather than like changing the existing file 2015-06-12 01:47:58 apps that already have the file open will not see the changes 2015-06-12 01:48:18 and the dev_t/ino_t change 2015-06-12 07:08:30 morning 2015-06-12 07:11:31 tdtrask: hey ted, you about? 2015-06-12 07:14:15 hello. i have an issue. i often have this error : mount: mounting /dev/mmcblk0p1 on /media/mmcblk0p1 failed: Resource busy 2015-06-12 07:14:33 have someone an idea ? 2015-06-12 07:17:14 Chipster1: is this when running apk update? 2015-06-12 07:18:07 ScrumpyJack yes 2015-06-12 07:18:13 or apk add something 2015-06-12 07:21:32 ScrumpyJack an idea ? 2015-06-12 07:27:50 isnt it only happening ones each boot? 2015-06-12 07:29:47 cache and lbu media is mounted ro, so each run of apk/lbu will remount it rw (and back) so cache and overlay can be writen. 2015-06-12 07:29:50 i don't know 2015-06-12 07:30:28 i will be careful 2015-06-12 07:30:43 it doesnt seem to break anything 2015-06-12 07:30:53 just give that message. 2015-06-12 07:31:07 after a succesfull remount, the message is gone. 2015-06-12 07:31:22 could be proc is not displaying mount correctly 2015-06-12 07:31:31 but after a remount it does 2015-06-12 07:31:39 ncopa: any idea about this? 2015-06-12 07:36:41 clandmeter: yeah, another apk update and it's fine 2015-06-12 07:39:45 seems /media/usb is not remounted ro again. 2015-06-12 07:44:51 mdocml is generating errors as well 2015-06-12 07:45:04 is that a known issue? 2015-06-12 07:56:53 I automates some things so it can pose concern 2015-06-12 07:59:47 looks like its only apk who has this issue. its probably not correctly mounting media back to readonly. 2015-06-12 08:00:22 ok 2015-06-12 08:07:32 clandmeter: i dont know 2015-06-12 08:07:45 ncopa: its caused by apk i think 2015-06-12 08:07:48 already explained to fabled 2015-06-12 08:07:59 i have seen at times that it is not able to remount back to ro 2015-06-12 08:08:07 its always now 2015-06-12 08:08:10 i thought it was due to the modloop mount 2015-06-12 08:08:18 atleast for my last 4 tmpfs installs 2015-06-12 08:08:53 it's on raspberry pi 2015-06-12 08:09:03 1 and 2 2015-06-12 08:10:10 oh 2015-06-12 08:10:17 i think i just fixed one related issue in apk 2015-06-12 08:10:35 i think apk leaked index file's fd 2015-06-12 08:10:43 and if it was open, remount ro might've failed 2015-06-12 08:11:46 ok 2015-06-12 08:30:31 anyone successfully built the nvidia linux graphics drivers under musl for alpine? 2015-06-12 08:37:15 ScrumpyJack: you mean nivdia proprietary driver? 2015-06-12 09:04:51 clandmeter: yes 2015-06-12 09:06:44 its a binary blob 2015-06-12 09:07:26 but there was some talk about it in #musl 2015-06-12 09:22:28 heh, i just read the backlog 2015-06-12 09:23:01 i like this community 2015-06-12 09:23:33 instead of just discussing which crypto lib to use, ideas are discussed how a good lib should be implemented 2015-06-12 09:24:18 and instead of just complaining about openssl ABI breakage, fabled suggests a fix: http://marc.info/?l=openssl-dev&m=143409903726582&w=2 2015-06-12 09:44:46 clandmeter: yikes - binary it is :( 2015-06-12 09:46:18 poo 2015-06-12 10:05:44 xf86-video-nouveau might save me 2015-06-12 10:06:06 if you dont play games, it should be enough. 2015-06-12 10:10:14 i need to drive 4 screens on the NVS 510 :( 2015-06-12 10:14:09 i didnt try more then 2 on intel. 2015-06-12 10:17:16 i have 3 working on intel (alpine linux) 2015-06-12 10:18:02 but managers want to buy some crap HP box with nvidia card 2015-06-12 10:18:38 i think the nouveau driver is pretty mature, but i didnt test it recently. 2015-06-12 10:22:44 i hope i never have to :) 2015-06-12 10:52:02 4 screens 2015-06-12 10:52:04 cool! 2015-06-12 13:59:55 ScrumpyJack: I was afk all day yesterday, now I'm about 2015-06-12 14:40:49 ahoi 2015-06-12 15:14:27 Hello, i'm new to alpine, i'm trying to start lxdm on boot, i already did that "rc-update add lxdm", but it doesn't work, any idea, please 2015-06-12 15:46:50 does lxdm start when you run rc-service lxdm start ? 2015-06-12 15:52:36 ScrumpyJack, yes 2015-06-12 15:52:50 i solved the issue, lxdm-lang was missing 2015-06-12 15:55:25 cool 2015-06-12 15:55:29 ScrumpyJack: hey 2015-06-12 15:55:36 you were looking for me? 2015-06-12 16:06:20 tdtrask: yes. i have a lxc contrainer with acf webinterface facing the outside world. it rebuilds itself every hour. could it be of any use to you to show off all the work you do on acf? maybe a link in the wiki somewhere quiet 2015-06-12 16:09:01 ACTION doesn't recommend exposing ACF to the outside world since it allows you to manage sensitive info 2015-06-12 16:09:07 like ssh keys and config 2015-06-12 16:09:18 how confident are you that noone can escape the sandbox? 2015-06-12 16:11:43 fairly confident. i could put together a repo that excludes the acfs for ssh and other sensitive ones 2015-06-12 16:12:38 ACTION is not opposed, but doesn't want to be blamed :D 2015-06-12 16:16:18 ok, i'll spin it up over the weekend and see where it's at 2015-06-12 16:19:06 great, thanks 2015-06-12 16:19:18 let me know when it's ready and I'll take a look 2015-06-12 18:17:18 Anybody here have X11 + GNOME or XFCE running on a virtual box install? 2015-06-12 18:25:39 xfce works very well under KVM 2015-06-12 18:26:49 http://it-offshore.co.uk/linux/alpine-linux/30-alpine-linux-spice-kvm-desktop 2015-06-12 18:27:01 I can't seem to get X11 to load. I'm not entirely sure why. it's complaining about /dev/fb0 and VESA, somethign about "cannot read int vect" 2015-06-12 18:31:02 have you run setup-xorg or what's it called? 2015-06-12 18:33:09 yeah, I ran that earlier without any issues 2015-06-12 18:34:31 it might be a bug in upstream x11 2015-06-12 18:35:02 if this ubuntu bug is the same problem I'm running into: https://bugs.launchpad.net/xserver-xorg-driver-vesa/+bug/1432899 2015-06-12 22:14:34 how to find $OSTYPE in sh? 2015-06-12 22:16:01 uname 2015-06-12 22:17:30 dalias: thanks! 2015-06-13 04:31:40 http://juniper.github.io/libxo/libxo-manual.html 2015-06-13 04:31:49 we needed a "why not freebsd?" argument? there is one 2015-06-13 04:32:00 :) 2015-06-13 04:33:01 it is their systemd 2015-06-13 04:34:35 in what way? 2015-06-13 04:34:43 redefining lots of base system stuff in a broken way? 2015-06-13 04:36:00 yes 2015-06-13 06:23:10 Anybody know anything about mysql on alpine? Looks like a dummy package and the wiki says to: /etc/init.d/mysql setup, except there's no mysql in init.d after apk add mysql mysql-client 2015-06-13 06:30:00 Maybe I should install it manually instead 2015-06-13 06:52:42 I guess I should see about getting it to work with MariaDB 2015-06-13 07:22:09 I guess it tries to use perl's DBI 2015-06-13 07:36:13 ? 2015-06-13 07:37:08 i'm pretty sure people are using one of the mysql variants tho i'm not sure which 2015-06-13 07:37:19 i try my best to avoid it :) 2015-06-13 07:39:42 yeah I'd prefer to avoid it 2015-06-13 09:51:01 kaniini: that lib looks nice 2015-06-13 09:51:30 wonder if it handles stuff in reverse as well 2015-06-13 10:27:41 good morning... I'm trying to bootstrap alpine, but can't get the apk signature verified 2015-06-13 10:27:47 my rough steps are: 2015-06-13 10:27:59 wget http://wiki.alpinelinux.org/cgi-bin/dl.cgi/v3.2/main/x86_64/apk-tools-static-2.6.1-r0.apk -O-|tar -zxvf- 2015-06-13 10:28:15 wget http://dl-4.alpinelinux.org/alpine/v3.2/main/x86_64/alpine-keys-1.1-r0.apk|tar -zxvf- 2015-06-13 10:28:25 openssl dgst -verify /etc/apk/keys/alpine-devel@lists.alpinelinux.org-4a6a0840.rsa.pub -signature sbin/apk.static.SIGN.RSA.alpine-devel@lists.alpinelinux.org-4a6a0840.rsa.pub sbin/apk.static 2015-06-13 10:28:30 Verification Failure 2015-06-13 10:29:23 of course the same outcome without etc/apk/keys/alpine-devel@lists.alpinelinux.org-4a6a0840.rsa.pub (relative path…) 2015-06-13 13:24:43 any tips on filesystem choice for an external hdd that's going to hold mostly video files (200M~2G each, there might me a couple ~10M files around too)? 2015-06-13 13:25:16 I'm mostly looking at NTFS, ext4 and NTFS 2015-06-13 13:26:49 Windows compat would be nice-to-have but it's not a deal-breaker, as it's probably going to be connected to a Linux file server 90% of the time 2015-06-13 13:37:23 if they're not > 4G FAT32 will give you the best compatibility. Two partitions? FAT32 and NTFS for >4G support? Those are the most common issues 2015-06-13 13:37:52 i try to forget about these types of things - I run everything ext4 here, but there's no other OS's around 2015-06-13 13:52:04 as I said, Windows is a minor issue 2015-06-13 13:52:24 And I REFUSE to put FAT on ANYTHING these days 2015-06-13 13:52:44 Sorry, but that time's gone 2015-06-13 13:53:15 It's bad enough Android forces it on you for SD cards -.- 2015-06-13 15:16:10 someone in here i think it was was talking about wireless usb adaptors 2015-06-13 15:16:30 athat were small and worked well in linux 2015-06-13 15:16:35 but i can't remember the link to it 2015-06-13 15:17:27 https://www.reddit.com/r/archlinux/comments/2dsa13/switching_from_os_x_broadcomwl_80211nx5ghz_support/cjsrr22 2015-06-13 15:17:32 i read this and it made me kind of sad 2015-06-13 15:17:43 i was sure someone linked this though https://www.thinkpenguin.com/gnu-linux/penguin-wireless-n-dual-band-usb-adapter-gnu-linux-tpe-nusbdb 2015-06-13 15:48:36 actually it might have been this one 2015-06-13 15:48:37 https://www.thinkpenguin.com/gnu-linux/penguin-wireless-n-usb-adapter-gnu-linux-tpe-n150usb 2015-06-13 15:52:47 sillysausage: made you sad? why? 2015-06-13 15:53:02 Orochimarufan: i seem to think it was you that was talking about it 2015-06-13 15:53:16 I was part of it, yeah 2015-06-13 15:53:18 well i have a macbook 10,1 2015-06-13 15:53:24 and it has a shitty broadcum in it 2015-06-13 15:53:30 [–]zman0900 2 points 9 months ago 2015-06-13 15:53:31 The bcm4331 in my mac is shit. Both the b43 and broadcom-wl drivers can't maintain a steady connection, and neither support 5 GHz. Even in OS X, the card won't connect to my 5 GHz network even though it can see it. I ended up getting one of these instead, and while it's a bit pricy and runs extremely hot, it is very reliable. 2015-06-13 15:53:34 particularly that reply 2015-06-13 15:53:53 the BCM4331 actually does support 5GHZ but it seems neither apple or linux drivers allow you to do that, which is interesting 2015-06-13 15:53:59 o.O 2015-06-13 15:54:13 so im considering a separate one, seems that ath9k is the only one that does not need a firmware blob 2015-06-13 15:54:23 https://www.thinkpenguin.com/gnu-linux/penguin-wireless-n-dual-band-usb-adapter-gnu-linux-tpe-nusbdb 2015-06-13 15:54:29 seems that's the only one that does 5ghz 2015-06-13 15:54:56 i was tempted to buy one of these https://www.thinkpenguin.com/gnu-linux/penguin-wireless-n-usb-adapter-w-external-antenna-gnu-linux-tpe-n150usbl 2015-06-13 15:55:00 for stealing the neighbor's wifi 2015-06-13 15:55:14 but it doesn't support 5GHZ 2015-06-13 15:55:45 https://www.thinkpenguin.com/gnu-linux/penguin-wireless-n-usb-adapter-gnu-linux-tpe-n150usb 2015-06-13 15:55:59 that one i think would be a bad idea, because i use it at uni, and im not sure where the access points are, they're hidden 2015-06-13 15:56:01 I use a rtl8812au 2015-06-13 15:56:11 it has 802.11ac 2015-06-13 15:56:19 ah yes 2015-06-13 15:56:27 which one? 2015-06-13 15:56:30 drivers aren't in mainline but opensource 2015-06-13 15:56:46 i think it was someone else that used that old one 2015-06-13 15:56:55 oh ok, well i can recompile my kernel 2015-06-13 15:57:06 is AC finalized yet? 2015-06-13 15:57:19 nah, you just compile the kernel module 2015-06-13 15:57:28 it's an out-of tree module 2015-06-13 15:57:48 right 2015-06-13 15:58:22 it's a minor inconvenience because you need to recompile the module after kernel upgrade, but other than that it's fine 2015-06-13 15:58:27 how are realtek, these days 2015-06-13 15:58:44 if it doesn't use blobs, it must be the first open source driver for an ac chipset 2015-06-13 15:58:54 While I don't have anything to compare against, it seems to work just fine 2015-06-13 15:59:12 Also, I don't use it for the AC, just 5GHz 802.11n 2015-06-13 15:59:21 Since our AP doesn't do AC (yet) 2015-06-13 15:59:50 yeah my university does actually have AC 2015-06-13 16:00:02 But I found the situation with linux-compatible 802.11a+n to be equally grim 2015-06-13 16:00:12 so I thought to myslef "Might aswell go all the way" 2015-06-13 16:00:14 which is why i sorta thought it would be a good idea considering the BCM4331 only does N 2015-06-13 16:00:21 yeah Orochimarufan same deal 2015-06-13 16:00:33 And it was only ~30€ on amazon 2015-06-13 16:00:49 i wonder if they are planning on merging the driver 2015-06-13 16:01:03 I think Realtek prefers having it out-of-tree 2015-06-13 16:01:18 which seems odd 2015-06-13 16:01:34 it seems to share source with their drivers for other platforms (it has a platform abstraction layer) 2015-06-13 16:01:50 Anyway, this is what I got: http://www.netis-systems.com/en/products/Dualband-Series/904.html 2015-06-13 16:02:29 I use a version of this https://github.com/Grawp/rtl8812au patched to work with linux 4.0 2015-06-13 16:02:32 https://aur.archlinux.org/packages/rtl8812au-git-dkms/ 2015-06-13 16:02:48 sounds good 2015-06-13 16:03:07 lol which git repository is realtek's 2015-06-13 16:03:18 I doubt they have a public one 2015-06-13 16:03:42 they ship the code to their customers 2015-06-13 16:04:07 like, there's a copy on the disk that comes with the netis one 2015-06-13 16:04:17 that's a neat idea 2015-06-13 16:04:34 how the usb dongle is the aerial 2015-06-13 16:04:43 you can also download it from most distributors 2015-06-13 16:05:07 yeah i use archlinux 2015-06-13 16:05:12 You probably want to keep this repo close too: https://github.com/pld-linux/rtl8812au 2015-06-13 16:05:13 i think that package in aur pulls it from git 2015-06-13 16:05:31 yeah there are plenty of people that have a copy on github 2015-06-13 16:05:42 https://aur.archlinux.org/packages/rtl8812au-git-dkms/ 2015-06-13 16:05:47 https://aur.archlinux.org/packages/rtl8812au_asus-dkms/ 2015-06-13 16:05:48 I'm probably have my own version up there 2015-06-13 16:05:49 i noticed two 2015-06-13 16:05:52 i wonder what the difference is 2015-06-13 16:06:06 dunno 2015-06-13 16:06:25 Description: rtl8812au linux kernel driver for ASUS USB-AC56 Wireless Dual-Band USB Adapter 2015-06-13 16:06:31 oh maybe that was some modified asus one 2015-06-13 16:06:40 i like that dongle Orochimarufan 2015-06-13 16:06:51 looks like it can have extra strong aerials attached 2015-06-13 16:06:51 Yeah I liked it too, hence I got it 2015-06-13 16:06:53 when you need them 2015-06-13 16:06:58 is that right? 2015-06-13 16:07:01 yeah, you can screw off the antennas 2015-06-13 16:07:11 nice. 2015-06-13 16:07:19 But in my case, that's not needed :) 2015-06-13 16:07:27 yeah 2015-06-13 16:07:36 there's a few wifi APs around where i live 2015-06-13 16:07:39 i should make use of them 2015-06-13 16:07:55 they all seem to have WPS enabled too 2015-06-13 16:08:11 I also like how they actually advertise Linux support :) More companies need to do that 2015-06-13 16:08:24 yes 2015-06-13 16:08:29 i think they deserve rewarding for that 2015-06-13 16:08:44 And while I've not tried it yet, it seems to have working Soft-AP too 2015-06-13 16:10:47 i better check if it supports monitor mode 2015-06-13 16:12:29 $ iw phy1 info 2015-06-13 16:12:31 ... 2015-06-13 16:12:35 Supported interface modes: 2015-06-13 16:12:35 https://forum.aircrack-ng.org/index.php?topic=709.0 2015-06-13 16:12:35 * IBSS 2015-06-13 16:12:35 * managed 2015-06-13 16:12:35 * AP 2015-06-13 16:12:36 * monitor 2015-06-13 16:12:37 * P2P-client 2015-06-13 16:12:38 * P2P-GO 2015-06-13 16:12:41 ... 2015-06-13 16:12:43 nice 2015-06-13 16:13:05 haven't tried anything fancy yet though 2015-06-13 16:13:35 the TPE-NUSBDB has: 2015-06-13 16:13:37 Master Mode (/w hostapd) 2015-06-13 16:13:39 Station Mode 2015-06-13 16:13:41 Monitor Mode 2015-06-13 16:13:43 AP Mode 2015-06-13 16:13:45 IBSS Mode 2015-06-13 16:13:47 Legacy (11g) operation 2015-06-13 16:13:49 HT support 2015-06-13 16:13:51 TX/RX 11n AMPDU aggregation 2015-06-13 16:13:53 HW Encryption 2015-06-13 16:13:55 Suspend/Resume 2015-06-13 16:13:57 which is most of those 2015-06-13 16:14:08 not sure what station mode is 2015-06-13 16:14:14 or master mode 2015-06-13 16:14:41 I also can't try out monitor right now cause I'm only ssh'd into that laptop right now 2015-06-13 16:14:52 messing with it would obviously kick me out :P 2015-06-13 16:16:17 I'll note that the Realtek RTL8812AU chipset seems ubiquitous in the current 802.11ac USB WiFi adapters on the market, from the most inexpensive models to the rather pricey ones. But, it appears monitor mode is a no-go in it. 2015-06-13 16:16:23 well that's a worry 2015-06-13 16:16:34 yeah it doesn't seem to actually work 2015-06-13 16:16:44 https://github.com/gnab/rtl8812au/issues/24 2015-06-13 16:17:42 OTOH, i'm not aware of another AC chipset that does it either 2015-06-13 16:17:58 yeah neither am i 2015-06-13 16:18:06 infact all the other AC ones are closed source too 2015-06-13 16:21:10 Used the 4.3.8 GPL driver from Realtek which has no monitor support. It's beyond me how Realtek can actually write GPL drivers, but can not be bothered to contribute a driver that follows the current mac80211 stack. In fact, I'm surprised that the OP got EDIMAX EW-7811UTC going as from my understanding RTL8811AU is basically the 1x1:1 version of RTL8812AU. 2015-06-13 16:21:16 that's probably why it's never going to be in the kernel 2015-06-13 16:22:06 i might hold off then 2015-06-13 16:23:04 That might not be a bad idea 2015-06-13 16:23:26 Though it remains to be seen whether there's going to be any improvement on the matter 2015-06-13 16:23:51 precisely. 2015-06-13 16:23:53 Network solution vendors don't seem to particularly care about Linux in a home environment 2015-06-13 16:24:14 (You wouldn't use a USB WiFi dongle in Rack servers, now would you?) 2015-06-13 16:24:35 well exactly, in any case there's no access points that support AC and do not support N 2015-06-13 16:24:42 which is why i think there's so few in that area 2015-06-13 16:25:09 While I agree that the realtek linux driver was probably an afterthought, it's an afterthought that works for me and I'm glad they did it 2015-06-13 16:25:26 Afterthoughts are still superior to ignorance ;) 2015-06-13 16:25:57 indeed 2015-06-13 16:26:08 but as i want to use monitor mode, that's not really suitable for me :) 2015-06-13 16:26:20 and if you really wanted, you could probably implement whatever new kernel interface they have on top of their HAL 2015-06-13 16:27:22 I could totally see myself doing that if the need came up, in fact :) 2015-06-13 16:28:22 That's the awesome thing about FOSS: instead of wasting energy complaining, you can just spend that energy fixing whatever you'd be complaining about. In the end, that's a win-win for everyone involved 2015-06-13 16:29:06 http://www.ieee802.org/11/Reports/802.11_Timelines.htm 2015-06-13 16:29:09 if you look there actually 2015-06-13 16:29:15 802.11AC isn't actually ANSI approved yet 2015-06-13 16:29:27 So what? :P 2015-06-13 16:29:33 apparently that's due in March 2016 2015-06-13 16:29:51 and i think a lot of chipsets were actually created prior to it even being finalized 2015-06-13 16:30:36 probably, but they seem to work just fine 2015-06-13 16:31:05 Final or Conditional 802 EC Approval 2015-06-13 16:31:20 Final 802.11 WG Approval 2015-06-13 16:31:23 that's the important one 2015-06-13 16:31:35 Nov 2015 2015-06-13 16:31:48 actually, what makes you say 802.11ac isn't ANSI approved? 2015-06-13 16:32:10 AFAICT, it's been approved 2013-12-11 2015-06-13 16:32:35 hmm 2015-06-13 16:32:35 IEEE Std P802.11ac-2013 2015-06-13 16:32:41 i might have looked at the wrong oone 2015-06-13 16:33:03 IEEE Std P802.11REVmc i think i might have been looking at 2015-06-13 16:33:09 yeah 2015-06-13 16:33:25 I guess that's just a new standards document 2015-06-13 16:33:36 yeah, so that's probably not the reason 2015-06-13 16:33:51 but non-kernel drivers are likely to never get much attention 2015-06-13 16:33:58 it's a pity the ath10k drivers require firmware blobs 2015-06-13 16:34:04 where as the ath9k ones did not 2015-06-13 16:34:30 companies tend to do blobs a lot lately :( 2015-06-13 16:34:53 yeah fucking shits me 2015-06-13 16:35:02 because then they don't even support the product properly 2015-06-13 16:39:12 although what's interesting 2015-06-13 16:39:17 is b43 is apparently labeled 2015-06-13 16:39:22 " 2015-06-13 16:39:25 b43 - An excellent and fully supported driver" 2015-06-13 16:39:32 according to this: https://wireless.wiki.kernel.org/en/developers/Documentation/mac80211 "Proper mac80211 stack drivers" can't even support 802.11ac yet 2015-06-13 16:39:43 Here is a quick review of the features supported in mac80211. 2015-06-13 16:39:43 IEEE 802.11abgn 2015-06-13 16:39:44 ah 2015-06-13 16:40:06 though the wiki might as well be out-dated 2015-06-13 16:41:00 oh well i might try out my b43 and if it doesn't work too well 2015-06-13 16:41:26 surely mac80211 is supporing 802.11ac by now 2015-06-13 16:41:28 buy that ath9 adaptor (TPE-NUSBDB) 2015-06-13 16:42:44 So basically, as I understand it, while most linux drivers for softmac devices use (and should use) the mac80211 interface to give the kernel control of the softmac part, while the realtek driver does the softmac work itself and uses the raw cfg80211 interface to tell the kernel it's a hardmac device 2015-06-13 16:45:50 I wonder how hard it would be to "out-source" the softmac work to the kernel 2015-06-13 16:46:19 it would probably compromise the HAL, which makes it understandable that realtek has no interest in doing it IMHO 2015-06-13 16:47:59 http://www.aircrack-ng.org/doku.php?id=compatibility_drivers was an interesting read 2015-06-13 16:50:14 https://github.com/lwfinger/rtlwifi_new/issues/9 also interesting 2015-06-13 16:50:31 and it seems I was right about realtek and mac80211 2015-06-13 16:50:44 so in short seems to be 1. Atheros (ath9k), 2. Broadcom (if you have support from b43, but note there won't be 5GHz), 3. Intel 2015-06-13 16:51:05 not all broadcom devices are supported by b43, particurly newer ones 2015-06-13 16:51:15 seems broadcom purposely removed monitor mode from their wl/sta drivers 2015-06-13 16:53:25 anyway, I was in the middle of doing something :O 2015-06-13 17:03:57 https://wireless.wiki.kernel.org/en/users/Drivers/brcm80211 2015-06-13 17:04:06 hmm seems to indicate that the BCM4331 was added 2015-06-13 17:04:28 https://wiki.archlinux.org/index.php/broadcom_wireless 2015-06-13 17:04:33 oh softmac 2015-06-13 17:42:29 anyone used this? http://douaneapp.com/ 2015-06-13 17:42:34 https://aur.archlinux.org/packages/?O=0&K=douane 2015-06-13 17:59:36 looks interesting 2015-06-13 19:02:27 linux seems to have it's issues with UAS :O 2015-06-13 19:03:53 yup... UAS driver is broken >.< 2015-06-13 19:04:56 not that I REALLY need UAS... 2015-06-13 19:07:24 Actually, this smells like the typical kernel-upgrade-got-rid-of-matching-kernel-modules-and-not-restarted-into-new-kernel-yet issue 2015-06-13 19:07:28 >.< 2015-06-13 19:10:14 theeere you go: 2015-06-13 19:10:16 [632610.940876] usbcore: registered new interface driver uas 2015-06-13 19:10:17 :) 2015-06-13 20:34:51 hello 2015-06-13 20:35:05 i have a small issue : 2015-06-13 20:35:27 i use abuild checksums and it return this error : 2015-06-13 20:35:28 /usr/bin/abuild: line 2045: can't create /var/cache/distfiles/0.1.tar.gz.lock: Permission denied 2015-06-13 20:35:39 have someone an idea ? 2015-06-13 20:38:45 you need to be root 2015-06-13 20:38:59 i try with root 2015-06-13 20:39:18 abuild checksums 2015-06-13 20:39:18 >>> ERROR: : Do not run abuild as root 2015-06-13 20:40:33 you need to set the permissions on the directory to abuild group 2015-06-13 20:40:42 ok 2015-06-13 20:41:38 http://wiki.alpinelinux.org/wiki/Include:Setup_your_system_and_account_for_building_packages 2015-06-13 20:41:50 thx 2015-06-13 20:41:53 i use these ones: 2015-06-13 20:42:01 sudo addgroup abuild sudo chgrp abuild /var/cache/distfiles sudo chmod g+w /var/cache/distfiles 2015-06-13 20:42:15 (almost at the bottom of that page) 2015-06-13 20:44:31 it's also useful to change $CLEANUP in /etc/abuild.conf to: 2015-06-13 20:44:31 #CLEANUP="srcdir pkgdir deps" 2015-06-13 20:44:31 CLEANUP="deps" 2015-06-13 20:44:44 so you can check things after a successful build 2015-06-13 20:45:22 ok 2015-06-14 01:16:50 When working on an APKBUILD file how do I prevent abuild from cleaning/deleting the fake root? 2015-06-14 01:23:34 nivardus: check abuild.conf 2015-06-14 01:31:28 I use the following in /etc/abuild.conf 2015-06-14 01:31:32 # what to clean up after a successful build 2015-06-14 01:31:32 #CLEANUP="srcdir pkgdir deps" 2015-06-14 01:31:32 CLEANUP="deps" 2015-06-14 01:31:57 brilliant, thanks @clandmeter @BitL0G1c 2015-06-14 10:07:44 http://www.pcworld.com/article/242938/spherical_flying_drone_looks_like_something_from_scifi_movies.html , nice!! 2015-06-14 10:13:33 vkrishn: go ahead 2015-06-14 10:13:48 dont ask and runaway 2015-06-14 10:13:50 ;) 2015-06-14 10:14:48 I do check the logs 2015-06-14 10:14:54 thanks 2015-06-14 10:19:49 vkrishn: logs of private chats? 2015-06-14 10:21:18 I could pop here, if said, "you around" 2015-06-14 10:22:00 I am thinking of adding znc on the box 2015-06-14 10:22:37 I think I would like to be on some php releated sites 2015-06-14 10:22:53 now that php 7alpha is out 2015-06-14 10:24:08 and they jumped php coding standard from psr from 4 to 7 2015-06-14 10:24:44 for a awile I thought I must have missed a lot 2015-06-14 10:26:04 vkrishn: why not use irssi+tmux? 2015-06-14 10:27:52 ok would try that 2015-06-14 10:29:22 does irssi support private chat ? 2015-06-14 10:30:27 im using it 2015-06-14 10:30:35 you mean query? 2015-06-14 10:36:11 :), yes, thanks 2015-06-14 13:32:23 just set irssi+tmux, with some irc channels and enabled logs... seems nice 2015-06-14 17:06:12 ehlo :) 2015-06-14 17:13:26 Has anyone (besides me) had issues starting samba from /etc/init.d/ when configured as a AD? I seem to be missing the ability to start samba from the script and now smbd and nmbd.. or am I missing something? 2015-06-14 17:13:38 err. 2015-06-14 17:13:54 s/now smbd/not smbd/ 2015-06-14 18:53:09 aight fuck it then. 2015-06-14 20:04:30 ACTION is back (gone 02:30:02) 2015-06-14 22:12:05 t0b1nux: could you not? 2015-06-15 02:05:14 greetings, I am dropping in to ask if anyone is able to help with an Xorg issue 2015-06-15 07:02:53 morning campers 2015-06-15 07:19:48 ScrumpyJack: evening :D 2015-06-15 07:32:05 are you not in Bulgaria systmkor ? 2015-06-15 07:40:00 no 2015-06-15 07:40:10 California 2015-06-15 07:43:54 ScrumpyJack: UTC -8 2015-06-15 07:47:04 :) 2015-06-15 07:49:16 ScrumpyJack: oh I got a sample APKBUILD from Justin C. http://bugs.alpinelinux.org/issues/3963 2015-06-15 07:51:08 I was wondering your thoughts on moving libseccomp stuff forward were 2015-06-15 07:53:51 ScrumpyJack: I would totally be down to join an Alpine Linux meetup but I don't have funds for traveling to Europe :( 2015-06-15 08:04:10 systmkor: yeah, that's a bummer 2015-06-15 08:04:37 where are we with libseccomp? 2015-06-15 08:08:17 hi there, I am having problem on my laptop, I added alpine-desktop after setup-xorg-base but when I startxfce4, mouse and keyboard hang 2015-06-15 08:08:30 I can only power off the machine, all else is ignored 2015-06-15 08:08:52 ScrumpyJack: I support the meetup endeavour if you guys need help with planning I can schedule free time 2015-06-15 08:09:23 mhe_: make sure you start dbus and udev before you startx 2015-06-15 08:09:41 mhe_: also make sure you have necessary drivers 2015-06-15 08:10:23 indeed. what graphics card is on your laptop? 2015-06-15 08:10:48 you could also try setup-box which will give you a nice Mate desktop (under devel) 2015-06-15 08:11:09 I added video-intel and fbdev, input-synaptics, evdev, keyboard, mouse 2015-06-15 08:11:56 startxfce4 shows a desktop and the choose default popup the first time xfce4 starts but I cannot click, type or move mouse 2015-06-15 08:12:42 rc-update add udev 2015-06-15 08:12:52 rc-service udev start 2015-06-15 08:13:24 same with dbus 2015-06-15 08:14:48 ok will try, on my desktop machine I did not do so and no problems there 2015-06-15 08:18:32 if you can't move the mouse, it sounds like dbus 2015-06-15 08:20:15 systmkor: your APKBUILD didn't work for me 2015-06-15 08:21:48 ah, hrmm I think that's Justin's but either way I'll put on my to-do list to make sure it works 2015-06-15 08:22:13 once it can compile, does there exist any tools for porting programs to comply with libseccomp 2015-06-15 08:22:41 basically scanning systemcalls for a program and white-list the ones used by a package/program 2015-06-15 08:22:54 so that can be partially automated 2015-06-15 08:27:20 hmm 2015-06-15 08:29:20 If not I've thought about building tools to help automate so it's easier to write correct libseccomp filters/configs & CAP information for grsec, etc. 2015-06-15 08:35:12 that's beyond my skill set :) 2015-06-15 08:35:46 libseccomp is in testing now 2015-06-15 08:40:28 hmm, it didn't build for me with Justin's APKBUILD file. I'll look later 2015-06-15 08:44:34 dumb question: why is php 5.6.10 not in the repositories although the corresponding APKBUILD has been updated from version 5.6.9 to 5.6.10 3 days ago? http://pkgs.alpinelinux.org/package/main/x86/php 2015-06-15 08:45:44 its edge 2015-06-15 08:45:49 you are probably on stable? 3.2 2015-06-15 08:51:35 mhe_: setup-xorg-base 2015-06-15 08:51:54 that was the first thing I tried 2015-06-15 08:52:05 tried both the 3.1.4 and 3.2 isos 2015-06-15 08:52:43 mhe_: so dbus was already started? 2015-06-15 08:52:52 did not check 2015-06-15 08:53:10 will try a reinstall when I get home 2015-06-15 08:53:20 mhe_: ah, i though you were in front of your laptop 2015-06-15 08:53:41 no, at the work office on a desktop 2015-06-15 08:53:47 try setup-box if you can 2015-06-15 08:59:46 mhe_: if your keyboard/mouse didn't work it was probably mdev, I think switching mdev to udev requires a reboot 2015-06-15 09:00:14 I did some reboots 2015-06-15 09:01:19 clandmeter: yes, I'm on stable - but how can you tell that 5.6.10 is only in edge? o_O 2015-06-15 09:23:17 try apk search -X repo php 2015-06-15 09:24:22 ScrumpyJack: 5.6.9 only 2015-06-15 09:25:04 I've also tried another mirror server (nl.alpinelinux.org) and of course I've done "apk update" an "apk upgrade -U", but none of that helped 2015-06-15 09:25:12 apk search -X http://liskamm.alpinelinux.uk/edge/main/ php 2015-06-15 09:25:30 http://liskamm.alpinelinux.uk/edge/main/x86_64/php-5.6.10-r0.apk 2015-06-15 09:25:35 well, at least openssl 1.0.2c got installed, but not php 5.6.10 2015-06-15 09:26:14 btw, is there some specific reason why an update like this is not pushed to the stable repo straight away? 2015-06-15 09:26:25 after all it fixes quite a bunch of security vulnerabilities 2015-06-15 09:27:09 btw, your command still only yields 5.6.9 2015-06-15 09:28:19 perhaps it would break any dependancies that some pacakges might have on php-5.6.9 2015-06-15 09:29:00 I'll just file a bug report so we'll see 2015-06-15 09:29:57 was libeventlog removed from 3.2? 2015-06-15 09:31:56 looks like it. it appears in 3.1 but not 3.2 2015-06-15 09:32:13 you could pull if from 3.1 specifically if you like 2015-06-15 09:42:36 that's what I do 2015-06-15 09:42:50 disappeared package looks weird, though 2015-06-15 09:42:56 i am working on backporting php fixes 2015-06-15 09:43:13 but not as weird as unblocking syslog with strace 2015-06-15 09:49:09 re libeventlog: http://dup.pw/aports/af9daae8df66 2015-06-15 09:49:32 looks like the source tarball went missing 2015-06-15 09:53:33 bobrik: you can file a bug for it 2015-06-15 09:53:39 i think we should try provide it for v3.2 2015-06-15 10:10:16 ncopa: sry for the dumb question, but why do you bother with backporting at all an not just ship the most recent version? I mean it's a scripting language, so I can hardly imagine just shipping 5.6.10 could break anything 2015-06-15 10:12:41 (don't wanna attack anyone, but I'm kinda new to Alpine and I couldn't find anything like "The Alpine Philosophy" in the wiki, so I don't know yet what is customary in the alpine community and where you do things differently then, let's say, the Archlinux people) 2015-06-15 10:18:16 btw, generally speaking, alpine seems to be exactly what I've been looking for for years 2015-06-15 10:22:14 grsec kernel by default, openrc instead of systemd, a decent build system (I just love (A)PKGBUILDs), a package management which doesn't drown in useless meta dependencies and artificial package splits, an optional web interface that looks reasonably secure and extensible to be used for embedded network devices, and even musl instead of glibc 2015-06-15 10:23:20 oh, and fairly up-to-date packages of course =) 2015-06-15 10:32:23 buerofalke: http://www.alpinelinux.org/about/ 2015-06-15 10:34:12 clandmeter: thx, I've read that of course, but I was more talking of something like this: https://wiki.archlinux.org/index.php/The_Arch_Way 2015-06-15 10:34:56 kiss is our main philosophy 2015-06-15 10:34:59 :) 2015-06-15 10:35:24 and security of course 2015-06-15 10:35:35 shoudl rename it to kisss 2015-06-15 10:36:52 and we dont like vendor lock-in 2015-06-15 10:40:10 but we do like lunch :) 2015-06-15 11:44:23 clandmeter: I like tmux+irssi, now I learning irssi script/module writing 2015-06-15 11:44:50 I really like to free myself of logging into the box to send msgs 2015-06-15 11:45:39 and I should be able to send/chk msgs from browser too ;) 2015-06-15 11:46:08 you can ssh from chrome ;-) 2015-06-15 11:47:02 so plan is an irssi module/script that checks ~/.irssi/freenode/channel/msg file every few seconds and publish it 2015-06-15 11:47:17 don't know if that is possible though 2015-06-15 11:47:57 now if thats possible, I only need a mean to publish onto that file using any protocol/means 2015-06-15 11:47:57 that sounds more like a bot to me 2015-06-15 11:48:40 with one diff, it does not respond to some or 2015-06-15 11:49:37 and talking to person just seems normal 2015-06-15 11:50:13 its more like a passive/proxy 2015-06-15 11:53:03 "ssh from chrome" , harr.. hrrr, I thought someone said kisss 2015-06-15 11:56:09 talking or irssi, how do you increase the scroll back buffer? 2015-06-15 11:56:12 and yes even send from normal smartphone having websocket browser 2015-06-15 11:56:39 (i'm at work and irssi site blocked and can't be fussed to spin up tsocks) 2015-06-15 12:17:49 found it 2015-06-15 13:53:11 hello 2015-06-15 13:54:07 <_mjones> hello. 2015-06-15 13:56:22 Yo folks. 2015-06-15 13:56:47 I tried installing Alpine this weekend, but the iso refused to boot. 2015-06-15 13:57:06 <_mjones> uefi boot or mbr? 2015-06-15 13:57:18 mbr 2015-06-15 13:57:26 i7 laptop, not an old one, 2011 2015-06-15 13:57:47 <_mjones> Tried the same iso in any other machines? 2015-06-15 13:58:08 not really... I only have this laptop 2015-06-15 13:58:15 it was a usb stick though 2015-06-15 13:58:24 but I used unetbootin 2015-06-15 13:58:49 <_mjones> Did you get unetbootin menu? 2015-06-15 13:58:57 and then tried using grub from my current linux installation, pointing to (hd1,1)/alpine...iso 2015-06-15 13:59:17 <_mjones> interesting fallback approach. 2015-06-15 13:59:19 nah, the error was about some syslinux thing 2015-06-15 13:59:27 basically it's the iso itself 2015-06-15 13:59:40 <_mjones> syslinux error from usb or from fallback approach? 2015-06-15 13:59:43 has something to go with my laptop (alienware m11x r3) 2015-06-15 13:59:47 both 2015-06-15 13:59:52 both approaches, same error 2015-06-15 14:00:12 <_mjones> the exact error message would be helpful. 2015-06-15 14:00:39 i dont think our iso is compatible like that 2015-06-15 14:00:40 I will try to get it again now 2015-06-15 14:01:08 it should be... why is it not a hybrid btw? 2015-06-15 14:01:18 its not hybrid 2015-06-15 14:01:19 I could just dd it to my flash drive then... 2015-06-15 14:01:43 you can copy it to usb, and isntall syslinux bootloader 2015-06-15 14:02:01 would be much simpler... I rarely see optical drives todays... and Alpine is especially good for embedded things 2015-06-15 14:02:33 voidnull, it's _not_ isohybrid 2015-06-15 14:02:37 <_mjones> @clandmeter: is the image not set up to be directly dd-able to usb? 2015-06-15 14:02:49 use setup-bootable to extract .iso to usb stick or similar 2015-06-15 14:02:50 _mjones: nope 2015-06-15 14:03:03 you need an hybrid iso for that. 2015-06-15 14:03:23 i dont remember the issue why we dont have it. 2015-06-15 14:03:27 maybe fabled does 2015-06-15 14:03:27 voidnull if you used unetbootin to make the usb stick, you'll need to replace a file on your us 2015-06-15 14:03:30 usb 2015-06-15 14:03:33 <_mjones> @clandmeter: I hadn't heard that term used for it before. I learned something. 2015-06-15 14:04:06 its very easy to make it bootable, so i dont see the need for all these tools 2015-06-15 14:05:01 clandmeter: agreed, setup-bootable works very well 2015-06-15 14:05:11 ScrumpyJack: replace which file? 2015-06-15 14:05:24 ScrumpyJack: well, you will need to have alpine booted already 2015-06-15 14:05:44 but wihtout alpine, you can copy the iso contants to usb (fat32) 2015-06-15 14:05:49 then syslinux -i.... 2015-06-15 14:06:27 any distro has syslinux, and windows has syslinux.exe (which also preps mbr) 2015-06-15 14:06:39 "SYSLINUX 4.03 2010-10-22 EDD Copyright (C) 1994-2010 H. Peter Anvin et al" 2015-06-15 14:07:04 that's what I get as the only output when I try to boot from the usb stick 2015-06-15 14:07:23 what's the difference between vanilla and minimal, btw? 2015-06-15 14:07:31 grsec 2015-06-15 14:07:39 vanilla has no grsec patches 2015-06-15 14:07:42 ah 2015-06-15 14:07:58 <_mjones> no pax either? 2015-06-15 14:08:15 plain vanilla 2015-06-15 14:08:33 like icecream but sweeter :) 2015-06-15 14:09:23 it has no patches whatsoever 2015-06-15 14:09:29 http://git.alpinelinux.org/cgit/aports/tree/main/linux-vanilla 2015-06-15 14:09:37 http://git.alpinelinux.org/cgit/aports/tree/main/linux-grsec 2015-06-15 14:09:38 voidnull: i can't remember. i put it on the wiki. let me check 2015-06-15 14:09:42 mennu32 i think 2015-06-15 14:09:45 <_mjones> The poster can't boot alpine so can't use setup-bootable. 2015-06-15 14:09:56 <_mjones> You have to admit there's a bootstrap problem. 2015-06-15 14:10:23 you dont need setup-bootlable 2015-06-15 14:10:46 but its handy when you have alpine installed already 2015-06-15 14:11:03 there was talk about adding images to our downloads 2015-06-15 14:11:05 so you can dd them 2015-06-15 14:12:05 voidnull: what happens if you remove all files from your usb, and copy all contents from iso to it? 2015-06-15 14:12:25 oh that will also remove syslinux... 2015-06-15 14:13:04 <_mjones> Yeah, I'm looking at setup-bootable now. 2015-06-15 14:13:35 <_mjones> But a hybrid image would be excellent, and easy to support... 2015-06-15 14:14:46 ncopa: what was preventing us from making our iso's hybrid? 2015-06-15 14:14:59 i remember something with fat32 2015-06-15 14:16:31 the main problem is init script 2015-06-15 14:16:50 i looked at it before v3.2 release 2015-06-15 14:17:46 what i want to do is look for boot repo and apkovl on any block device found by blkid 2015-06-15 14:21:07 voidnull you will need to copy the mboot.c32 file suppled by your distro into /boot/syslinux/ on the USB stick 2015-06-15 14:24:10 clandmeter: will try that in a few, thank you; ScrumpyJack you mean to copy mboot.c32 from .iso? or from my currently installed linux distro? that's just stange that I need to do so much voodoo, Alpine seems line a solid distro, and it's hard to install :( 2015-06-15 14:24:49 and unetbootin usually makes everything boot, even for uefi, iirc... weird 2015-06-15 14:25:19 voidnull: which host os are you using to create the usb? 2015-06-15 14:25:31 windows 7 2015-06-15 14:25:46 <_mjones> Is this on the wiki? I usually only boot VMs. 2015-06-15 14:25:53 sorry for using that word around here 2015-06-15 14:26:12 but it's my work computer, so I used it to reinstall my OS 2015-06-15 14:26:18 voidnull: do you have syslinux.exe? 2015-06-15 14:26:59 I could get it... but I also have Void Linux installed on my laptop right now, I want to switch to Alpine 2015-06-15 14:27:32 so I could use that Void linux thing and run syslinux from it 2015-06-15 14:28:21 but I thought that was what unetbootin does: installs syslinux and unpacks the iso 2015-06-15 14:28:26 http://nl.alpinelinux.org/alpine/v3.2/main/x86_64/syslinux-6.03-r1.apk <- has syslinux.exe 2015-06-15 14:28:35 you can rename it to tar.gz 2015-06-15 14:28:41 thanks 2015-06-15 14:28:54 <_mjones> voidnull: I admit that Alpine isn't the easiest to install for all use-cases, but if you want to see a nonautomated install that requires skill, check out the instructions for Arch Linux... 2015-06-15 14:29:00 I have to do something now, will return here later, thank you guys 2015-06-15 14:29:36 syslinux.exe -i -m -a driverletter: 2015-06-15 14:30:10 you so that on an usb disk which is fat32 and has the contents of our iso 2015-06-15 14:40:10 I found the easiest way to make a usb was to mount it inside an alpine kvm guest 2015-06-15 14:40:25 in libvirt 2015-06-15 14:43:44 libvirt settings http://hastebin.com/huqaxicunu.xml - the hexcodes are found with lsusb 2015-06-15 20:19:53 hi everyone. I'm having a problem setting up xorg-base, it just freezes for hours on installing encodings. Any help would be great :/ 2015-06-15 20:25:05 ant0k: hours? 2015-06-15 20:25:09 yes 2015-06-15 20:25:19 http repo?" 2015-06-15 20:25:49 mm, haven't tried changing the 1st one 2015-06-15 20:26:29 nl.alpinelinux.org 2015-06-15 20:26:36 for hours sounds like a network issue 2015-06-15 20:26:46 any logs or dmesg? 2015-06-15 20:27:01 is it package isntalling or post install scripts? 2015-06-15 20:27:07 sounds ok, gonna try nother ones then 2015-06-15 20:27:11 im running setup-xorg-base 2015-06-15 20:27:22 its just a meta pkg 2015-06-15 20:27:22 or setup-gparted-desktop, which runs the previous also 2015-06-15 20:28:07 it gets stuck on 10-11% when reaching encodings, that's it. My first install, sorry for flood 2015-06-15 20:29:34 sounds strange 2015-06-15 20:29:44 can you install encodings first? 2015-06-15 20:30:58 already did, it installed fine with apk add encodings, but still it gets stuck when running xorg setup script 2015-06-15 20:32:17 maybe it's my router problem (wlan), gonna try it with eth and write back 2015-06-15 20:53:36 Hello all, I was looking for information on alpine arm ports and specifically any builds for the beaglebone black. Last reference was in March 2014 here: http://lists.alpinelinux.org/alpine-devel/3611.html Can anyone point me in the right direction? 2015-06-15 20:57:32 baetheus: hi 2015-06-15 20:57:40 clandmeter Hello 2015-06-15 20:58:31 did you try images from http://dev.alpinelinux.org/~tteras/ ? 2015-06-15 20:58:38 there are some uboot ones 2015-06-15 20:59:02 I haven't, but I'll check that out today, thanks. 2015-06-15 20:59:37 clandmeter: sorry again, tried doing apk add encodings once again, actually it doesn't gets installed (with apk add), getting stuck on 95% every time, so i guess it's not a network prob 2015-06-15 20:59:54 baetheus: hi 2015-06-15 21:01:35 encodings 1.0.3-r1, tried changing repo to de/yandex/etc, running apk with -Uf. Anyways, thanks for your time 2015-06-15 21:03:54 ant0k: Hello 2015-06-15 21:09:00 ant0k: can you download the apk from the mirror? 2015-06-15 21:09:45 i guess its http://nl.alpinelinux.org/alpine/v3.2/main/x86_64/encodings-1.0.3-r1.apk 2015-06-15 21:09:55 if you're on 64b 2015-06-15 23:07:16 Hey clandmeter, I'm taking a look at alpine-uboot-150424-armhf from that link you supplied. Looks like tteras or whoever was nice enough to provide an MLO and u-boot.img for the beaglebone black, but I'm unsure if that u-boot is going to look for /boot on the ssd card or the internal 2gb mmc. Do you have any idea? 2015-06-15 23:08:55 Err, sorry, not ssd, microsd. 2015-06-15 23:22:12 hi guys, that package contains snmpwalk and snmpget? 2015-06-15 23:23:45 ready, sorry, net-snmp-tools :-D 2015-06-16 04:40:28 update from the guy who's trying to install Alpine from a USB drive 2015-06-16 04:40:51 followed the instructions about syslinux, managed to get to the boot: line 2015-06-16 04:41:28 after I press return, the OpenRC ( I think that's what it's called) freezes at "Loading hardware drivers" 2015-06-16 04:41:54 tried disabling as many hw devices as I could in BIOS, but that didn't help 2015-06-16 04:45:18 I've used a mini image, now going to try the full (300mb) iso 2015-06-16 04:48:28 nah, same story :( 2015-06-16 04:51:36 will try vanilla 2015-06-16 04:51:38 have you run linux on this machine before? 2015-06-16 04:51:50 yeah, for 3 years 2015-06-16 04:52:18 it sounds like a kernel issue (either main kernel or modules) not being compatible with your hardware 2015-06-16 04:52:42 it's the first time in 7 years (4 own laptops/netbooks, more than 20 various distros) that I have so many issues during the install 2015-06-16 04:53:12 the kernel is 3.18... I used 3.14 before, 4.0.4_1 from Void Linux works fine on it 2015-06-16 04:53:45 <__number5__> do you have standalone nvidia or amd video cards 2015-06-16 04:53:46 <__number5__> ? 2015-06-16 04:54:10 I have optimus, intel and nvidia cards 2015-06-16 04:54:22 it might be helpful to pastebin "lspci -v" from a working linux boot on the machine 2015-06-16 04:54:24 <__number5__> that might be the issue 2015-06-16 04:54:36 multiple video cards in the machine? :) 2015-06-16 04:55:22 that sounds like it could be a source of breakage if they're not initialized in a way so as not to stomp over each other 2015-06-16 04:55:24 <__number5__> the default kernel might not have the right drivers\ 2015-06-16 04:55:31 but i've never used such a setup so i don't know 2015-06-16 04:55:48 http://forum.alpinelinux.org/forum/installation/pc-freeze-during-boot 2015-06-16 04:56:10 it's two cards, i915 is the default driver for that, should not cause problems :( 2015-06-16 04:56:29 frustrating, I really want to have a musl distro and I think Alpine is the best case for that 2015-06-16 04:56:47 I didn't like Void that much, but at least it was easy to install it 2015-06-16 04:58:02 <__number5__> voidnull: have you tried install alpine with a cdrom? 2015-06-16 04:58:13 __number5__, i really don't think that would help 2015-06-16 04:58:22 the problem is drivers crashing the kernel 2015-06-16 04:58:49 the stuff to suppress mode setting on the kernel command line might not work when the drivers are loaded later as modules... 2015-06-16 04:58:51 <__number5__> yep, you are right, it already passed the kernel loading phase 2015-06-16 04:58:53 (just a guess) 2015-06-16 04:59:23 if you could temporarily remove all but one vid card that might be a good way to diagnose the problem (see whether that fixes it) 2015-06-16 04:59:23 the first distro to behave like that on this laptop 2015-06-16 04:59:30 oh it's a laptop :( 2015-06-16 04:59:34 ye 2015-06-16 04:59:36 then it's probably hard to remove one 2015-06-16 04:59:42 I could try 2015-06-16 04:59:51 let me find my soldering iron 2015-06-16 04:59:53 when you get the first ash prompt... 2015-06-16 05:00:18 I only get the "boot:" thing, then I press return and then it freezes on the hardware drivers 2015-06-16 05:00:37 i thought on the forum you could boot to single-user or whatever 2015-06-16 05:01:37 if so, try rm'ing the kernel modules for video hardware before resuming the boot 2015-06-16 05:01:40 and see if that fixes it 2015-06-16 05:06:22 clandmeter, the images of 3.2 supersede the ones in ~tteras 2015-06-16 05:08:36 too tired to mess with it now, it's 1am... wish the installation was easier... if I fail with it tomorrow night, will probably stay with Void Linux 2015-06-16 05:08:42 thanks for the tips, g'night 2015-06-16 07:20:29 hello there. i just installed alpine in a virtualbox. however, i can't find the apk-tool anywhere. can i somehow install that afterwards? 2015-06-16 07:23:33 du5tball: hi. you should have 'apk' command? 2015-06-16 07:23:38 its not there at all? 2015-06-16 07:24:10 that's the exact problem. it simply doesnt exist. find / -name apk yields three folders (in lib, etc, and i forgot where the third was.) 2015-06-16 07:25:51 ... the third reinstall solved the problem. weird. 2015-06-16 07:28:16 while i'm here anyways: does alpine have an all-encompassing webfrontend (other than webmin)? 2015-06-16 07:28:52 i think so but i haven't used it... 2015-06-16 07:31:34 would you know it's name? 2015-06-16 07:32:09 oh, it says so on the wiki-page about webmin 2015-06-16 07:34:00 i forget -- what's it called? 2015-06-16 07:34:08 ACF 2015-06-16 07:35:52 ah yes 2015-06-16 07:44:35 why is this thing torturing me during installation >.< 2015-06-16 07:48:05 this is like trying to install a hackintosh... if you run the installer several times, you finally get a complete system 2015-06-16 07:51:59 du5tball: do you mean setup-alpine? 2015-06-16 07:52:11 yup 2015-06-16 07:52:51 it doesn't handle typos and mistakes very well. if you make a mistake with your answers, it's best to reboot the install media i find. 2015-06-16 07:54:10 the point is, i didn't make typos. i run the thing, everything works fine until it starts partitioning and installing. partitioning works, but at some point it fails to run apk and actually copy programs to the disk 2015-06-16 07:54:33 hmm. any error message? 2015-06-16 07:54:52 one, that i didn't save. and no, i will not run the installer again :D it works now 2015-06-16 07:57:05 Welcome to Alpine Linux! It's all about fiddling and tweaking and adjusting, which you've already experienced :) 2015-06-16 07:57:56 thank god my main system is archlinux 2015-06-16 07:58:30 would be nice to know what failed 2015-06-16 07:58:34 so it can be fixed 2015-06-16 08:01:07 du5tball: you won't get "would be nice to know what failed so it can be fixed" from the lead developer of archlinux ;P 2015-06-16 08:02:58 just running a second instance. during mirror testing, i get "ERROR: http://liskamm.alpinelinux.uk/edge/main: Protocol error". the system determines dl-7.alpinelinux.org as the fastest 2015-06-16 08:03:29 next error: "Updating repository indexes... ERROR: http://dl-7.alpinelinux.org/alpine/v3.2/main: Protocol error" 2015-06-16 08:03:54 and then it fails? 2015-06-16 08:05:41 nah, keeps going. during installation itself however it says "ERROR: apk-tools-2.6.1-r0: Protocol error". then it generates the initramfs and the next error comes along: "/sbin/apk: No such file or directory. Skipped". it goes on telling me blocks, "/boot is device /dev/sda1" and gives me back the console 2015-06-16 08:06:04 it fails to install apk 2015-06-16 08:06:22 "protocol error" sounds like network failed for some reason 2015-06-16 08:06:41 is dl-7 mirror ok? 2015-06-16 08:06:43 ok, there are 2 problems 2015-06-16 08:06:53 it's only on those three instances though. everything else goes on fine and the system is bootable 2015-06-16 08:07:00 problem 1: some of the mirrors are not ok 2015-06-16 08:07:15 problem 2: install of some of the packages fails, in this case it seems to be apk-tools 2015-06-16 08:07:25 leading to a bootable system, but without apk-tools 2015-06-16 08:07:42 it should have bailed out at that point instead of just continue 2015-06-16 08:07:43 out of 10 installations, 8 wound up not installing apk-tools 2015-06-16 08:12:19 can i see somewhere which mirror apk chose? after installing, i mean 2015-06-16 08:12:30 in /etc/apk/repositories 2015-06-16 08:12:41 you can edit it manually too 2015-06-16 08:13:07 ok. i think i know why http://liskamm.alpinelinux.uk fails 2015-06-16 08:14:47 this works: apk update --quiet --repository http://liskamm.alpinelinux.uk/edge/main --repositories-file /dev/null; echo $? 2015-06-16 08:14:59 this does not: apk update --quiet --repository http://liskamm.alpinelinux.uk//edge/main --repositories-file /dev/null; echo $? 2015-06-16 08:17:27 ncopa: liskamm is fine though right? it's that thing with updated-since in the HTTP request no? 2015-06-16 08:17:44 liskamm appears to be fine 2015-06-16 08:17:47 i'm trying dl-5.alpinelinux.org/alpine/v3.2/main as the mirror right now... everything i try to apk add fails now 2015-06-16 08:18:02 ncopa: phew :) 2015-06-16 08:18:13 i thought it was that http req thingy, but problem seems to be that it does not like double // 2015-06-16 08:18:16 in path 2015-06-16 08:18:28 du5tball: can you try: apk update? 2015-06-16 08:18:54 will do 2015-06-16 08:19:21 apk will download a package database. if the package database does not correspond with existing packages, it will give errors 2015-06-16 08:19:28 i foten do: apk add -U ... 2015-06-16 08:19:33 often* 2015-06-16 08:19:53 the -U will run an 'apk update' before trying to install 2015-06-16 08:20:46 nah, apk update fails. "temporary error. try again later" 2015-06-16 08:21:15 drat. back in a min. 2015-06-16 08:26:07 isn't there some option in bios to disable addon graphics card ? 2015-06-16 08:28:07 ah, ok you already tried 2015-06-16 08:31:34 half my system went belly up for no apparent reason. but noooo, work must use windows. 2015-06-16 08:36:06 du5tball: i filed a bug for one of the issues you had: http://bugs.alpinelinux.org/issues/4365 2015-06-16 08:36:27 with target 3.2.1 so i dont forget it before the 3.2.1 release 2015-06-16 09:21:24 how can I tell apk to install from aports head (want to add a new package in testing)? 2015-06-16 09:28:38 <_ikke_> justincormack: You can use pinning 2015-06-16 09:28:51 <_ikke_> you have to add the correct repository 2015-06-16 09:29:45 ncopa: thanks. sorry, had an insta-meeting >_> 2015-06-16 09:29:57 np 2015-06-16 09:32:02 which is the correct repo? Using apk -X git://git.alpinelinux.org/aports doesnt seem to work - trying to test http://git.alpinelinux.org/cgit/aports/commit/?id=9ca764d6b9a7b7e47625dd4e153943fffb499d28 2015-06-16 09:32:50 <_ikke_> Aports just contains the build files, not packages 2015-06-16 09:33:02 ah ok 2015-06-16 09:33:05 /edge/testing 2015-06-16 09:33:46 echo "@testing http://dl-7.alpinelinux.org/alpine/edge/testin" >> /etc/apk/repositories 2015-06-16 09:34:07 apk add -U libseccomp@testing 2015-06-16 09:34:08 ah thanks! 2015-06-16 09:34:23 i had a typo up there so don just copy paste 2015-06-16 09:34:24 apk -X http://nl.alpinelinux.org/alpine/edge/testing add libseccomp worked 2015-06-16 09:34:39 should work yes 2015-06-16 09:34:57 (after update) 2015-06-16 17:58:15 morning! 2015-06-16 17:59:52 I have a patch for a port in main, may I submit it here. 2015-06-16 18:00:01 More like a request feature in fact 2015-06-16 18:00:17 http://sprunge.us/ZSZa 2015-06-16 18:03:33 I don't understand why single precision have been enabled here. But the "default" is double precision, which creates libfftw3.so rather than libfftw3f.so 2015-06-16 18:04:10 I'm not quite sure which soft depends on fftw3, but it's most likely that they will depend on the double precision lib 2015-06-16 18:06:00 there are only 4 ports depending on it: testing/rosegarde, testing/liblastfm and two "unmaintained" ones 2015-06-16 18:08:20 i suspect there may be a desire to have both libs 2015-06-16 18:16:19 that would be an option indeed 2015-06-16 18:16:35 the thing is, it looks like you can't compile both at the same time 2015-06-16 18:16:57 Archlinux's take on this was to copy the initial directory 3 times 2015-06-16 18:17:09 and compile the libs 3 times, each time with different flags 2015-06-16 18:17:30 I must admit that I'm not fond of this approach, hence the patch I submitted 2015-06-16 18:18:16 but as long as I can have libfftw3.so and watch my music spectrum on the CLI, I'm fine :) 2015-06-16 18:18:35 (yeah, it's definitely NOT a critical issue) 2015-06-16 22:02:12 hi 2015-06-16 22:02:21 i have the following problem with mosh: 2015-06-16 22:02:31 Error relocating /usr/bin/mosh-server: _ZNK6google8protobuf11MessageLite17SerializeToStringEPSs: symbol not found 2015-06-16 22:02:56 about 20 lines like this 2015-06-16 22:03:07 same with mosh-client 2015-06-16 22:03:13 i'm using the latest version 2015-06-16 22:03:15 any idea? 2015-06-17 06:18:39 Nightwolf: it could be due to new libstdc++ ABI in gcc5 2015-06-17 06:18:52 i rebuilt protobuf 2015-06-17 06:18:58 seems like i need rebuild mosh too 2015-06-17 07:01:10 Fonts have changed on my twm desktop. Did i read somewhere that a font package got updated? 2015-06-17 07:10:02 yup 2015-06-17 07:10:23 we got new freetype version and added some patches to it 2015-06-17 07:36:09 ncopa: Can you rebuild geekie? http://sprunge.us/VOQD 2015-06-17 07:36:28 humpf 2015-06-17 07:36:29 ok 2015-06-17 07:36:45 :) thanks. sorry to be a pain 2015-06-17 07:37:00 libexiv2.so.13 2015-06-17 07:37:01 no 2015-06-17 07:37:06 its gcc5'd fault 2015-06-17 07:37:19 we need rebuild everything linked to libexiv2.so.13 2015-06-17 07:37:51 hum 2015-06-17 07:37:57 might be its enough to rebuild gexiv2 2015-06-17 07:39:50 no 2015-06-17 08:03:00 owncloud sync client is broken with the recent mesa update i think 2015-06-17 08:03:04 MESA-LOADER: could not create udev device for fd 15 2015-06-17 08:33:49 for me the synaptics touchpad driver fails at x startup (i have to rmmod/modprobe the i2c_hid module to fix it) 2015-06-17 08:34:32 (since i last updated alpine edge about a month ago) 2015-06-17 08:37:37 hm there is a hid_rmi module that should be able to drive the touchpad too.. may be i should try that 2015-06-17 09:13:47 ncopa: thanks, it works again :) 2015-06-17 09:14:10 thanks for reporting it 2015-06-17 10:51:09 is there any X11 xfontselect tool type thing already in the repos? 2015-06-17 12:25:22 ScrumpyJack: do you like new font rendering? 2015-06-17 14:20:18 is there any other way than a chroot install to get only a / and a swap 2015-06-17 14:20:45 I just did a chroot-install and can't get my eth0 up 2015-06-17 14:22:17 LXC container ? 2015-06-17 14:22:43 no, used the chroot-install wiki 2015-06-17 14:23:36 added a kernel and added networking with rc-update 2015-06-17 14:23:37 did you run 'setup-interfaces' ? 2015-06-17 14:23:54 if I run that it just lists manual config [no] 2015-06-17 14:24:09 it does not bring up the usual stuff 2015-06-17 14:24:16 I do have lo as interface 2015-06-17 14:25:48 on a new install be it lxc usb or kvm - I run setup-interfaces x 3 for dhcp & ifup eth0 2015-06-17 14:28:26 I used this http://wiki.alpinelinux.org/wiki/Installing_Alpine_Linux_in_a_chroot 2015-06-17 14:28:58 if you only need a build environment - LXC is much easier http://wiki.alpinelinux.org/wiki/LXC#Alpine_Template 2015-06-17 14:30:13 no it is not a build environment 2015-06-17 14:30:35 a stripped down install without a seperate /boot 2015-06-17 14:33:06 perhaps you need to setup a bridge for networking 2015-06-17 14:33:28 with bridge-utils 2015-06-17 14:34:48 & setup /etc/network/interfaces on the HOST something like http://hastebin.com/uvuniyexos.vala 2015-06-17 14:35:27 then in the guest just configure a static ip address on the same subnet 2015-06-17 14:36:32 somehing like this http://hastebin.com/xutipinehe.hs 2015-06-17 14:37:32 I have the first 2 lines you got and added auto lo 2015-06-17 14:37:39 auto eth0 I mean 2015-06-17 14:37:57 what is the host os ? 2015-06-17 14:38:06 it boots as alpine linux 2015-06-17 14:38:29 I used a chroot to setup and install, but I can boot 2015-06-17 14:40:11 so you have an alpine chroot inside alpine linux ? 2015-06-17 14:49:11 no I used a livecd to set up the chroot 2015-06-17 14:49:33 added everything I needed to boot and booted to the installation 2015-06-17 14:51:11 ok you don't need a bridge then - did you run setup-interfaces ? 2015-06-17 14:56:14 yes but it does not do what I expect 2015-06-17 14:57:06 it asks only if I want to add any manual config 2015-06-17 14:57:47 you have to 3 times for dhcp 2015-06-17 15:00:11 I worked with the setup-ifaces before 2015-06-17 15:00:23 normally first question is which interfaces do you want to ... 2015-06-17 15:00:40 but this time it goes straight to do you want any manual config 2015-06-17 15:00:48 so if I enter, the script is over 2015-06-17 15:03:34 I must have missed something with the installation 2015-06-17 15:12:40 mhe: ip link 2015-06-17 15:13:16 If eth0 doesn't show up there, check dmesg to see why your driver isn't loading 2015-06-17 15:14:10 If it does, try setup-interfaces -h to see if there's an option to pass it in as an argument 2015-06-17 15:14:28 I'm traveling without a computer currently or I'd be more specific 2015-06-17 15:21:03 I was missing a runlevel 2015-06-17 15:21:15 added udev and some other 2015-06-17 15:21:29 now it works as expected 2015-06-17 15:48:28 hi, I want to test alpine. But after creating a bootable usb stick I get: 2015-06-17 15:48:32 /sbin/init not found in new root. Launching emergency recovery shell 2015-06-17 15:48:34 Type exit to continue boot. 2015-06-17 15:48:40 any idea? 2015-06-17 15:52:58 <_ikke_> ferdinand: How did you create the stick? 2015-06-17 15:54:29 like described here: http://wiki.alpinelinux.org/wiki/Create_a_Bootable_USB 2015-06-17 16:02:13 I do not know if it matters, but host system is openSUSE 2015-06-17 16:30:55 ferdinand: it shouldn't matter, which iso did you use? 2015-06-17 16:36:42 alpine-3.2.0-x86_64.iso 2015-06-17 16:42:37 Is init in /sbin? 2015-06-17 16:43:09 Make sure the initramfs is on the stick 2015-06-17 16:43:50 Obviously if any step of the USB setup failed, you won't be able to boot 2015-06-17 16:44:30 init is just a symlink to /bin/busybox.\ 2015-06-17 16:45:04 But yeah, setup-boot must have gone wrong somewhere. 2015-06-17 16:53:13 I have checked the stick multiple times, all files are there. 2015-06-17 17:15:20 hmm... it does not work, even with unetbootin... I will try next release. 2015-06-17 18:23:41 Ferdinand - best way to make a USB stick is from inside another Alpine - mount a USB stick inside a KVM Alpine 2015-06-17 18:25:48 usb from usb! 2015-06-17 22:27:41 BitL0G1c: http://dev.alpinelinux.org/~clandmeter/alpine-mini-3.2.0-x86_64-512M.img.gz 2015-06-17 22:48:18 clandmeter do you want me to try & make a usb with it ? 2015-06-17 23:21:49 you can now login to your fwknop protected ssh from your phone http://www.cipherdyne.org/blog/2015/06/new-android-single-packet-authorization-client-fwknop2.html 2015-06-18 00:11:14 hey guys, a quick question: is hybernate/suspend working on alpine linux? 2015-06-18 05:24:18 hello 2015-06-18 05:24:28 i have some issue 2015-06-18 05:24:51 i make an apkbuild with abuild and it's done 2015-06-18 05:25:28 now i try to do apk update and after apk add my_apk and i have this error : 2015-06-18 05:25:34 ERROR: unsatisfiable constraints: 2015-06-18 05:25:34 sshd_autoban (missing): 2015-06-18 05:25:34 required by: world[sshd_autoban] 2015-06-18 05:25:43 have someone an idea ? 2015-06-18 05:31:49 might be out-of-date mirror, try running update again after you clear the package index, see http://wiki.alpinelinux.org/wiki/Alpine_Linux_package_management#Update_the_Package_list 2015-06-18 05:41:42 same error 2015-06-18 05:48:57 Diftraku it's done 2015-06-18 05:49:20 i wasn't in apk's directory 2015-06-18 09:18:53 Chipster1: make sure your repositories include your packages directory 2015-06-18 09:38:34 any i3wm users here? 2015-06-18 11:45:02 hey :) is suspend/hybernate working in apline linux for lenovo x230i laptops? 2015-06-18 11:59:12 i've had trouble getting it to work. what window manager? 2015-06-18 12:00:41 ScrumpyJack, xfce 2015-06-18 12:01:21 i am currently not on alpine linux, but i concider switching to it because it like the small and lightweight approach. but if hybernate is not working at all, then i have to look for something else 2015-06-18 12:03:18 uhm... which tool is used for user management? oO 2015-06-18 12:03:48 ScrumpyJack, do you have a lenovo x230i? 2015-06-18 12:04:19 scrumpyjack: i'm using i3, but on a different distro. 2015-06-18 12:08:33 leo-unglaub: no, i've got some HP thingy. suspend/hyber is pretty much generic these days. i know people have it working. mine doesn't but i haven't sat down to fix it 2015-06-18 12:09:07 leo-unglaub: suspend was/is working fine for me with xfce 2015-06-18 12:09:22 ScrumpyJack, okay, i see 2015-06-18 12:09:35 clandmeter, okay, then i will give Alpine a try today :) 2015-06-18 12:09:41 thanks to both of you! 2015-06-18 12:09:41 clandmeter: how does xfce call suspend? 2015-06-18 12:09:56 thats one part i didnt find out yet 2015-06-18 12:10:01 its greyed out for me 2015-06-18 12:10:06 so i use cmdline 2015-06-18 12:10:20 pm-suspend? 2015-06-18 12:10:22 clandmeter, on lid close would be enought for me 2015-06-18 12:10:36 ScrumpyJack: yes 2015-06-18 12:11:12 leo-unglaub: for lid close, acpi will notice and will trigger a configurable even 2015-06-18 12:11:41 leo-unglaub: ahills was going write about it. 2015-06-18 12:12:05 okay, great ! 2015-06-18 12:12:12 here is mine http://sprunge.us/cTgQ 2015-06-18 12:12:24 and do you know about wifi on a lenovo x230i? 2015-06-18 12:12:26 is that working? 2015-06-18 12:12:34 mosez: how do you start i3 2015-06-18 12:12:43 ScrumpyJack, that is the one from your wiki, right? 2015-06-18 12:13:04 i can't remember. what's your wifi chipset? 2015-06-18 12:13:29 scrumpyjack: in the past with lightdm and now with gdm 2015-06-18 12:13:58 so with something in ~/.xinitrc? 2015-06-18 12:14:14 ScrumpyJack, Centrino Advanced-N 6205 [Taylor Peak] 2015-06-18 12:14:39 https://github.com/tboerger/homeshick-suse/blob/master/home/.xinitrc yeah, but something like a default for opensuse. 2015-06-18 12:16:39 hmm, i'm using slim and i3, but i3 just starts the bar and mouse is in "waiting" - no kb input possible 2015-06-18 12:16:57 got no luck with slim on opensuse :) 2015-06-18 12:18:26 i don't think slim is the problem 2015-06-18 12:18:31 :( 2015-06-18 12:19:05 https://github.com/tboerger/homeshick-suse/tree/master/home/.i3 got to extend my i3 config 8) 2015-06-18 12:24:22 leo-unglaub: WiFi should definitely work, hibernate is disabled in grsec kernels by default (but I'm working on it), suspend (including lid) works perfect 2015-06-18 12:24:40 I'm on an x1 carbon, very similar to yours 2015-06-18 12:25:26 Currently I'm away though, if you want hibernation in the meantime I recommend using the vanilla kernel instead 2015-06-18 12:26:53 ahills, awesome! do i need the firmware package for wifi? or is my chipset supported with a free driver without a blob? 2015-06-18 12:27:07 I recommend pm-utils, or use the edge repo and grab my uswsusp package (read about repo pinning and fall in love with alpine a little more) 2015-06-18 12:27:34 leo-unglaub: Intel is always firmware but iirc it just worked with alpine 2015-06-18 12:27:49 okay, i will give it a try :) 2015-06-18 12:28:00 Your main concern will be picking a network manager 2015-06-18 12:28:03 as soon as the download has finished, thanks for all your help! 2015-06-18 12:28:21 ahills, no, i have my own network manager :) 2015-06-18 12:28:39 As much as I hate to stay it NetworkManager is to most feature-complete solution right now 2015-06-18 12:28:39 i forked the gnome network manager back in the days when it used gtk2 and was awesome! 2015-06-18 12:28:51 just to make sure, hibernation is disabled in grsec, you need to undisable it in the grsec patch and recompile, suspend should work oob 2015-06-18 12:29:07 Does your fork track the new CLI improvements? 2015-06-18 12:29:32 ahills, i completely split the package into a cli and a gtk version 2015-06-18 12:30:16 stf: I'm unable to resume from full disk encryption so far 2015-06-18 12:30:21 scrympyjack:if you have problems with mouse/keyb you might want to run setup-xorg-base, and make sure the udev daemon is correctly setup. 2015-06-18 12:30:26 ahills: you are? damn. 2015-06-18 12:30:50 hilite still w\o/rks ;) 2015-06-18 12:31:15 \o/ 2015-06-18 12:32:43 i should play with this myself. 2015-06-18 12:36:06 bCQc0P3DIqbh, did suspend with for you with encrypted harddrives? 2015-06-18 12:36:22 was just hybernate affected? 2015-06-18 12:41:05 hds are irrelevant for suspend, suspend goes to ram. 2015-06-18 12:42:47 leo-unglaub: suspend works perfectly out of the box, just grab pm-utils 2015-06-18 12:43:14 ahills, i ma still encryting my discs to install it on 2015-06-18 12:43:18 i will try it later on 2015-06-18 12:44:02 OK, enjoy 2015-06-18 12:45:01 i hope i will :) 2015-06-18 12:46:09 ahills: i guess you dont use xfce right? 2015-06-18 12:46:25 Nope 2015-06-18 12:46:49 How could your wm get in the way of suspend? 2015-06-18 12:49:05 its greyed out, and i dont know why. 2015-06-18 12:49:19 i need root to use pm-* 2015-06-18 12:49:23 maybe thts the case 2015-06-18 12:49:56 Get uswsusp and use s2ram 2015-06-18 12:50:24 Then when it works great move it to main ;) 2015-06-18 12:50:52 it would be nice to use the menu's in xfce, did you add support for it? :) 2015-06-18 12:51:10 at least on *buntu, you need to enable hibernation with polkit 2015-06-18 12:51:47 I don't know a single thing about xfce, but I imagine it's XML, so I likely won't touch it 2015-06-18 12:52:21 pm-utils looks for uswsusp though 2015-06-18 12:52:29 ya 2015-06-18 12:52:37 Maybe it JustWorks™? 2015-06-18 12:52:52 try pm-suspend or w/e it is as root 2015-06-18 12:54:40 whats the diff between them? i never used uswsusp 2015-06-18 12:56:05 *shrugs*, pm-utils is the go to thing on ubuntu, probably just pulls all depencies (if any) 2015-06-18 13:00:31 clandmeter: uswsusp is user space 2015-06-18 13:00:50 Some "new" kernel feature 2015-06-18 13:01:28 There's a uswsusp.txt or so 2015-06-18 13:07:03 ahills: ok got it. i thought it was something you made. i remember somebody made some daemon and aded it to aports. i think i got confused. 2015-06-18 13:21:22 bCQc0P3DIqbh: I have several wm working with this install of Alpine Linux, just not i3 :( 2015-06-18 13:23:16 leo-unglaub-is-a: what wm did you have in mind to use for AL? 2015-06-18 13:29:15 ScrumpyJack, xfce4 and i3 2015-06-18 13:29:36 depending on my location. laptop only i3, laptop with exterman screen xfce4 2015-06-18 13:37:58 clandmeter: I also did that, but not uswsusp 2015-06-18 13:47:38 there are spam messages on forums :( 2015-06-18 13:56:42 again? 2015-06-18 13:58:58 could be because i put the captcha words in one of the posts 2015-06-18 13:59:12 i think ill just remove those 2015-06-18 14:00:13 AmatCoder: i think i deleted those spam msg's 2015-06-18 14:03:13 alacerda: do we have the Mate desktop manager? 2015-06-18 14:03:19 clandmeter: I think they are still here 2015-06-18 14:03:45 AmatCoder: uri? 2015-06-18 14:03:56 is it caja? 2015-06-18 14:04:23 ScrumpyJack, yes 2015-06-18 14:04:24 Caja is the official file manager for the MATE desktop 2015-06-18 14:04:57 hmm. sorry, not caja. i meant "desktop manager" 2015-06-18 14:05:04 AmatCoder: ok i see it, its when logged out 2015-06-18 14:05:10 need to fix those permissions 2015-06-18 14:05:13 but i have to go 2015-06-18 14:05:24 in Mate, try System -> Preferences -> Desktop Preferences 2015-06-18 14:05:26 no ist gone 2015-06-18 14:05:30 refresh is the key 2015-06-18 14:06:03 i should probably int cache clean on a ban. 2015-06-18 14:06:10 init 2015-06-18 14:06:24 clandmeter: yes, now they are gone 2015-06-18 14:06:41 clandmeter: thanks :) 2015-06-18 14:07:03 ScrumpyJack, unfortunatly i dont have a MATE desktop here right now. 2015-06-18 14:07:10 :) 2015-06-18 14:07:23 My notebook is at home 2015-06-18 14:20:32 is there an iso file chat can be dd onto a usb drive? 2015-06-18 14:21:48 i mean dd if=alpine-linux.iso of=usb-device 2015-06-18 14:23:29 <_ikke_> Not provided by alpine linux 2015-06-18 14:23:46 <_ikke_> You can clone a usb device though when you created one 2015-06-18 14:29:20 <_ikke_> foobar_: The problem is that the iso is targeted to optical media, and they didn't make it hybrid 2015-06-18 14:30:28 is it hard to make it hybrid? 2015-06-18 14:30:38 <_ikke_> I don't know 2015-06-18 14:30:45 because my computer does not have a optical drive 2015-06-18 14:31:04 <_ikke_> foobar_: You can use something like unetbootin to create a bootable usb 2015-06-18 14:54:31 this is how i use to create my bootable usbs ^^ 2015-06-18 14:56:36 <_ikke_> Once you have one, you can dd that to an image 2015-06-18 15:02:55 vkris: snownews. what was the problem? can you remind me? 2015-06-18 15:39:36 _ikke_: i tryed unetbooin 2015-06-18 15:39:42 but it did not work :( 2015-06-18 15:47:41 the best way to create a USB stick is to use an alpine KVM 2015-06-18 15:47:51 & mount the usb in the kvm guest 2015-06-18 15:48:53 you can mount a usb stick in kvm with libvirt via something like http://hastebin.com/qoduzopeku.xml 2015-06-18 15:51:02 lsusb will tell you the correct vendor & product codes 2015-06-18 15:51:44 put ... looks like getting alpine on the usb drive is even harder than installing it 2015-06-18 15:52:25 I tried unetbootin & everything else - the only thing that worked for me was using an existing alpine installation 2015-06-18 15:53:18 BitL0G1c: do you still have the usb image? 2015-06-18 15:53:25 could you uplaod that one somewhere? 2015-06-18 15:54:37 I can try & make a new one - give me a little while 2015-06-18 15:55:24 BitL0G1c: that would be nice, because i only have a windows laptop from a co worker here and its hard to get it done on that device 2015-06-18 16:00:09 ok no problem - give me about an hour 2015-06-18 16:00:18 sure, thanks a lot! 2015-06-18 16:00:38 ;-) 2015-06-18 16:12:14 foobar_: http://dev.alpinelinux.org/~clandmeter/alpine-mini-3.2.0-x86_64-512M.img.gz 2015-06-18 16:13:32 BitL0G1c: i gave you that link for this reason ;-) 2015-06-18 16:14:07 clandmeter: awesome! 2015-06-18 16:14:27 clandmeter: can i dd that image on my usb drive? 2015-06-18 16:14:33 ah ok 2015-06-18 16:14:53 foobar_: yes 2015-06-18 16:17:35 foobar_: did you get an error with unetbootin? 2015-06-18 16:19:43 ScrumpyJack: no, it was just freezing up on boot 2015-06-18 16:19:51 i am trying the link from clandmeter now 2015-06-18 16:28:51 clandmeter: thanks, your link works fine! 2015-06-18 16:29:10 yw 2015-06-18 16:29:35 alpine will be made compatible with hubrid 2015-06-18 16:29:37 hybrid 2015-06-18 16:29:43 but probably 3.3 2015-06-18 16:30:36 clandmeter: awesome! 2015-06-18 16:31:17 now i just have to get my full disc encryption working :) 2015-06-18 23:43:18 foobar_: fde is easy, setup-disk can be given a /dev/mapper device, just make sure you set the kernel command line in update-extlinux.conf 2015-06-19 01:16:46 hey all :D. does anyone know how to run the generic arm Alpine image in qemu? 2015-06-19 01:17:09 I've tried putting it in a FAT32 partition in a raw disk image 2015-06-19 01:17:20 however I keep getting the qemu instance crashing 2015-06-19 06:34:58 hello 2015-06-19 07:01:38 good morning guys, i am trying to install alpine on an encrypted drive. i have prepared the luks encrypted partitions and they are now opened in /dev/mapper/, how can i pass them to setup-disk? or how do i continure from here? 2015-06-19 07:02:47 do i have to mount them somewhere and then start setup-apline? 2015-06-19 07:02:57 or do i have to use setup-disk? 2015-06-19 07:03:07 the wiki is not so clear on that 2015-06-19 07:31:24 i think you can pass a mount point to setup-disk 2015-06-19 07:32:13 mount your enc dev in /mnt then, for a complete install setup-disk -m sys /mnt 2015-06-19 07:33:54 ScrumpyJack: i will give it a try, what is recommended for /boot, ext4? 2015-06-19 10:39:57 is the person who write this script http://it-offshore.co.uk/linux/alpine-linux/25-alpine-linux-luks-encrypted-installations here in the channel? 2015-06-19 10:54:03 foobar_: BitL0G1c 2015-06-19 10:54:23 yes 2015-06-19 10:57:04 awesome :) 2015-06-19 10:57:35 BitL0G1c: a quick question, in your video you use setup-lvm .... that is now setup-partitons, right? 2015-06-19 10:57:43 yes 2015-06-19 10:58:08 it can setup any partition scheme (not just LVM) 2015-06-19 10:58:22 so i changed the name to avoid confusion 2015-06-19 10:58:28 okay, and in the current version there is a step for "disk layout" 2015-06-19 10:58:44 for only luks without lvm i need none? 2015-06-19 10:58:57 lvm | just a boot partition | or none (manual) 2015-06-19 10:59:09 choose "boot" 2015-06-19 11:02:45 BitL0G1c: hmm, the script still fails: "mount: wrong fs tyoe, bad option or bad superblock" 2015-06-19 11:02:49 type is ext4 2015-06-19 11:03:43 did you choose the mount points -- then "q" to quit ? 2015-06-19 11:04:11 yes 2015-06-19 11:05:25 ./dev/mapper/root -> /mnt 2015-06-19 11:06:27 can you paste the output on http://hastebin.com/ please 2015-06-19 11:07:03 BitL0G1c: i am trying it in a virtualbox let me try to ssh into that thing 2015-06-19 11:14:37 BitL0G1c: http://hastebin.com/hiyihejixe.vhdl 2015-06-19 11:15:26 ok - you need to 'apk add btrfs' 2015-06-19 11:15:43 BitL0G1c: i have it installed 2015-06-19 11:15:53 manually ding mkfs.btrfs works fine 2015-06-19 11:16:30 i think the problem is 2015-06-19 11:16:32 formatting /dev/sda1 as ext4 The file /dev/sda1 does not exist and no size was specified. 2015-06-19 11:16:41 how can sda1 not exist? 2015-06-19 11:17:38 foobar_: because there is no partition on sda ? 2015-06-19 11:17:53 I've seen that sometimes - rebooting the cd fixed it 2015-06-19 11:17:58 Coolgeek: no, there are partitions 2015-06-19 11:18:15 check in /dev - it won't be there 2015-06-19 11:18:26 but will be in fdisk -l 2015-06-19 11:18:40 never worked out why it happened 2015-06-19 11:19:32 for /boot you only really need 100m - I've never had any space problems with that size 2015-06-19 11:20:12 BitL0G1c: maybe the problem is that partprobe is not available and therefor the partiton table cannot be re-read 2015-06-19 11:20:24 foobar_: do you have partprobe ? 2015-06-19 11:21:14 or do : "echo 1 > /sys/block/sda/device/rescan" 2015-06-19 11:21:17 should work 2015-06-19 11:22:07 Coolgeek: partprobe and the echo dont work 2015-06-19 11:22:15 i think i have to reboot that vm 2015-06-19 11:22:30 yes is what fixed it for me 2015-06-19 11:22:49 i'll try a btrfs install too 2015-06-19 11:22:56 foobar_: you did that with user root ? 2015-06-19 11:23:07 Coolgeek: yes 2015-06-19 11:23:58 hdparm -z /dev/sda 2015-06-19 11:24:00 or 2015-06-19 11:24:05 sfdisk -R /dev/sda 2015-06-19 11:24:08 I think 2015-06-19 11:24:11 http://serverfault.com/questions/36038/reread-partition-table-without-rebooting 2015-06-19 11:28:45 BitL0G1c: http://hastebin.com/larupozuhu.hs 2015-06-19 11:28:51 it still does not work as expected 2015-06-19 11:32:40 apk add btrfs-progs & remove '-q' on line 279 - give me 5 minutes to update github 2015-06-19 11:34:17 BitL0G1c: i fixed it locally and after that fix it works fine! 2015-06-19 11:34:23 BitL0G1c: thanks for that amszing script! 2015-06-19 11:34:26 its awesome! 2015-06-19 11:34:48 ;-) thx for testing - will add the fixes now 2015-06-19 11:38:47 BitL0G1c: just an idea that currently comes to mind, it would be awesome to be able to select the luksFormat options as well 2015-06-19 11:39:07 key length, hash algorythm, .. 2015-06-19 11:40:19 yes - am planning to add that - if you choose say 'root' - it gives you the option to wipe or reconfigure LUKS 2015-06-19 11:40:38 just haven't added reconfigure yet 2015-06-19 11:41:19 (i.e choose the device mapper name) 2015-06-19 11:42:09 LUKS uses sensible defaults with a 5 second wait time to make brute forcing take forever 2015-06-19 11:43:03 BitL0G1c: http://i.imgur.com/AxTftXQ.png 2015-06-19 11:43:15 see create_crypt() around line 343 2015-06-19 11:43:47 he is not finding the keyfile for sda6 with would be /dev/mapper/tmp 2015-06-19 11:44:19 root 2015-06-19 11:45:18 ok - let me try with btrfs 2015-06-19 11:45:50 luksDump on sda6 is only showing one slot used 2015-06-19 11:46:03 but with the normal passphrase and the keyfile it should be two slots 2015-06-19 11:46:13 i think adding the new keyfile somehow failed 2015-06-19 11:47:32 what does your /etc/crypttab look like ? 2015-06-19 11:48:49 tmp UUID="...." /etc/luks-keys/tmp luks 2015-06-19 11:52:42 weird - mine mounted ok 2015-06-19 11:55:01 very strange 2015-06-19 11:55:13 i will look into it more tonight 2015-06-19 11:55:18 see you later! 2015-06-19 11:56:20 http://imgur.com/h4yT9w8 2015-06-19 11:56:27 cu 2015-06-19 11:56:49 BitL0G1c: i am going to add the key manually, then it should work as well 2015-06-19 11:57:18 setting up luks is not the hard part, but your script is soo helpful for that bootloader aund initramfs crap that i always get wrong *g* 2015-06-19 11:58:45 it also sets swap / lvm partition types some people miss when installing LVM too (causing a fail to boot) 2015-06-19 12:00:41 BitL0G1c: it works now in my virtualbox, i am trying in on my real computer ... if i dont come back ... it failed *g* 2015-06-19 12:00:59 hahaha ok 2015-06-19 13:45:06 ScrumpyJack: ok with issue I had with snownews, found atom2rss script :) 2015-06-19 13:46:44 but think would go for newsbeuter even if it does do columns display quite ok 2015-06-19 13:48:37 I mean does not* 2015-06-19 14:46:40 BitL0G1c: ping 2015-06-19 14:54:55 cool, xterm in testing 2015-06-19 14:55:13 thanks to clandmeter for that 2015-06-19 14:55:20 :) 2015-06-19 14:55:27 ;) 2015-06-19 14:58:05 I am going to reboot to testing it properly... 2015-06-19 15:04:37 you need to reboot to test xterm? :) 2015-06-19 15:07:54 ehy is the script http://it-offshore.co.uk/linux/alpine-linux/25-alpine-linux-luks-encrypted-installations working in wirtualbox, but not on my normal computer? 2015-06-19 15:13:17 fooobar - here - everything ok ? 2015-06-19 15:13:27 BitL0G1c: not really :( 2015-06-19 15:13:36 damn - what failed ? 2015-06-19 15:13:36 thr script fails on my normal computer 2015-06-19 15:13:56 ./setup-disk -E is not finding anything 2015-06-19 15:14:03 it just exits 2015-06-19 15:17:29 did you try ./setup-disk -E /mnt 2015-06-19 15:17:51 yes, i did to. if asks me for my root and then says: No boot media found 2015-06-19 15:18:06 i tryed mbr and gpt 2015-06-19 15:18:58 were you choosing "none" again & not "boot" for the partition scheme in setup-partitions ? 2015-06-19 15:19:19 i choose boot 2015-06-19 15:19:29 as you told me to 2015-06-19 15:22:26 BitL0G1c: now i wiped the entie disk and after "select LUKS ROOT device Mapper" i get the following 2015-06-19 15:22:38 No filesystem mounted @ /mnt 2015-06-19 15:23:05 but according to mount /dev/mapper/root is mounted on /mnt 2015-06-19 15:31:07 what does: awk "\$2 == \"/mnt\" { print \$1 }" /proc/mounts | tail -n 1 2015-06-19 15:32:00 give you ? 2015-06-19 15:32:54 it gives: /dev/mapper/root 2015-06-19 15:43:07 BitL0G1c: any ideas why this is not working? 2015-06-19 15:44:09 I can only really suggest checking /boot is correctly configured & everything is correctly mounted 2015-06-19 15:48:46 this code starts around line 1127 after 'if [ -n "$CRYPTROOT" ]; then' - perhaps after CRYPTROOT & CRYPTDISK add 'read -p "cryptroot=$CRYPTROOT cryptdisk=$CRYPTDISK" test' - to see if these are correctly detected 2015-06-19 15:51:05 BitL0G1c: i am going to wipe the entire disc and start another do over 2015-06-19 15:51:08 maybe this helps 2015-06-19 15:52:28 ok 2015-06-19 15:56:11 BitL0G1c: do you recommend gpt or dos? 2015-06-19 16:02:40 they both work - i use gpt now 2015-06-19 16:09:01 BitL0G1c: No boot Device detected. 2015-06-19 16:11:29 you need to work out why find_boot_partition() is not finding the boot partition 2015-06-19 16:13:30 BitL0G1c: the value that gets testet when it fails is /dev/sdb 2015-06-19 16:13:32 it 2015-06-19 16:13:42 it's probing /dev/sdb 2015-06-19 16:16:28 if this is for GPT it is running 'sgdisk /dev/sdb -A 1:show' 2015-06-19 16:16:37 then 'sgdisk /dev/sdb -A 2:show' 2015-06-19 16:17:00 & so on until it finds "legacy BIOS bootable" 2015-06-19 16:19:21 BitL0G1c: if sgdisk $1 -A $i:show |grep "legacy BIOS bootable" 1>/dev/null; then 2015-06-19 16:19:26 this line is never true 2015-06-19 16:19:45 it is if a partition is marked bootable 2015-06-19 16:20:13 BitL0G1c: the entire command is never returning any output 2015-06-19 16:20:39 yes because of 1>/dev/null 2015-06-19 16:20:59 no, i mean just sgdisk $1 -A $i:show 2015-06-19 16:21:04 so grep can never hit anything 2015-06-19 16:23:17 BitL0G1c: doen't it need "-A list" 2015-06-19 16:23:42 so you get the list containing "legacy BIOS bootable" ? 2015-06-19 16:24:27 it works for me http://hastebin.com/natiweyune.avrasm 2015-06-19 16:24:57 well that's odd 2015-06-19 16:24:59 hmm, very interresting 2015-06-19 16:25:35 I seem to be segfaulting out of autoconf-generated configure consistently on edge iff $SHELL is correctly pointed at zsh 2015-06-19 16:26:05 that's one hell of an odd condition 2015-06-19 16:26:18 BitL0G1c: how did you create that partition 2015-06-19 16:26:27 mybe i missed something there 2015-06-19 16:26:35 fooobar - setup-partition marks the GPT boot partition as bootable - if you are creating your own boot partition - you need to mark it as bootable 2015-06-19 16:27:10 BitL0G1c: i used your tool and i tryed it manually, both times the same 2015-06-19 16:30:00 the script would have run: sgdisk -A 2:set:2 /dev/sdb 2015-06-19 16:31:59 if you run that 2015-06-19 16:32:02 & then 2015-06-19 16:32:14 sgdisk /dev/sdb -A 2:show |grep "legacy BIOS bootable" 2015-06-19 16:33:35 you should see an output & setup-disk will stop quitting due to no boot partition 2015-06-19 16:34:12 BitL0G1c: my boot partition is /dev/sdb1, should i change ot to sgdisk -A 1:set1 /dev/sdb? 2015-06-19 16:36:03 the first partition is BIOS boot - partition 2 needs to be marked as bootable http://hastebin.com/pinarapucu.rib 2015-06-19 16:38:28 your parition type for partiton 2 is different, i dont have it set to EFI 2015-06-19 16:39:32 well this is the problem - just use the script to create the GPT boot partition 2015-06-19 16:40:14 MBR is a bit more forgiving if you want a bigger /boot - but really boot does not need to be more than 100meg 2015-06-19 16:44:44 wah, now the script does not find the new partition ... 2015-06-19 16:44:49 i have to reboot again *g* 2015-06-19 16:44:58 thats propobly the 100 time i wget your script today 2015-06-19 16:45:00 hahaha 2015-06-19 16:52:05 BitL0G1c: WAAAAAAAAAAAAA 2015-06-19 16:52:10 i found the fucking error 2015-06-19 16:52:57 ./setup-disk -E /mnt/ <- if you use TAB to get /mnt you get the last / as well and the script dies with the last / 2015-06-19 16:53:05 without the / it works fine 2015-06-19 16:53:44 ./setup-disk -E /mnt/ <- bad 2015-06-19 16:53:50 ./setup-disk -E /mnt <- good :) 2015-06-19 16:54:04 maybe you could add an info or trim the last / *g* 2015-06-19 16:54:19 so that the next guy who uses it does not spend 6 hours on that *g* 2015-06-19 16:56:29 didn't realise you were trying to setup-disk -E /mnt/ - yes can check for that 2015-06-19 16:57:10 it's the damn tab completion *g* 2015-06-19 17:05:41 hi. i'm new. i'm trying to compile libdvdcss and configure says the c compiler cannot create executables. 2015-06-19 17:07:47 apk add alpine-sdk 2015-06-19 17:08:30 Thanks, I thought as much. was trying various "base" type keywords 2015-06-19 17:11:03 if you are making APKBUILD's it is helpful to edit /etc/abuild.conf to something like http://hastebin.com/haruxogafi.vala 2015-06-19 17:11:16 so you can check things afterwards 2015-06-19 17:14:10 BitL0G1c: lol, it's still not over *g* i finished your script and rebooted .. it shows the boot screen but before he asks me for a passphrase i get the message: "Command required device and mapped name as arguments" 2015-06-19 17:14:14 any ideas on that one? 2015-06-19 17:19:26 BitL0G1c: ah, damn ... during boot /dev/sdb is called sda 2015-06-19 17:19:33 so cryptsetup fails 2015-06-19 17:21:38 the script sets UUID's in /etc/crypttab for this very reason 2015-06-19 17:23:32 I'll test with 2 disks 2015-06-19 17:26:23 BitL0G1c: yes, in crypttab it uses uuids ... but not in the /etc/update-extlinux.conf 2015-06-19 17:28:01 ok let me fix that 2015-06-19 17:29:49 thanks, dude you are awesome! 2015-06-19 17:41:47 will be a bit later - dinner first 2015-06-19 17:42:34 is somebody is hammering nl.a.o rsync? 2015-06-19 17:43:16 -is 2015-06-19 18:29:34 has anyone an idea why my keyboard worked the first time i started X but since a reboot my keyboard does not work under X anymore. The mouse never worked 2015-06-19 18:30:31 i have xf86-input-evdev installed 2015-06-19 18:31:53 i have also xf86-input-keyboard and xf86-input-mouse installed 2015-06-19 18:35:11 <_ikke_> how are you running x? 2015-06-19 18:36:26 _ikke_: startxfce4 2015-06-19 18:44:57 i hate X 2015-06-19 18:45:14 why the hell is my keyboard not working, when it worked after the first start 2015-06-19 18:47:15 <_ikke_> nothing in Xorg.0.log? 2015-06-19 18:47:48 _ikke_: nothing, X boots up nicely 2015-06-19 18:47:56 i also checked X outout via SSH 2015-06-19 18:56:42 hmm, the problem seams to be that udevd is not running/crashing 2015-06-19 19:31:25 anyone playing with Alpine ARM generic version with Qemu? 2015-06-20 03:30:56 hm can dropbear handle ed25519 keypairs 2015-06-20 07:02:18 ammunta: dropbear in server mode on my alpine box doesn't handle them at all. 2015-06-20 12:57:40 Is ca-certificates broken? 2015-06-20 12:57:45 WARNING: ca-certificates.crt does not contain exactly one certificate or CRL: skipping 2015-06-20 13:06:50 wCPO - i think that warning is normal rather than an error 2015-06-20 20:03:44 compiling gcc in Alpine on an RPi, this is going to take a while 2015-06-20 20:54:51 Hello all, I've been playing around with alpine-uboot-150424-armhf.tar.gz from dev.alpinelinux.org/~tteras and I've been unsuccessful in getting it running on a beaglebone black. If anyone has any experience, hints, or docs to help me out I'd be grateful. 2015-06-20 21:09:31 baetheus - ^^^stateless is probably the best person to speak to for that 2015-06-20 21:42:19 stateless: ping 2015-06-20 21:42:38 well i see freenode is busy sucking today 2015-06-20 21:43:05 :) 2015-06-20 21:43:39 freenode really needs to disable that stupid "Cannot change nickname while banned on channel" thing during DoS 2015-06-20 21:43:43 ACTION wonders who is ddosing it today 2015-06-20 21:44:16 or just disable it en 2015-06-20 21:44:21 or just disable it entirely since it's not effective 2015-06-20 21:44:33 oh, it's super effective against 14 year old children 2015-06-20 21:44:37 the same spamming using your nick can be achieved by just rapidly joining and parting with different nicks 2015-06-20 21:44:43 that's true 2015-06-20 21:45:12 and it makes it impossible to recover your nick after ghosting 2015-06-20 21:45:31 without either manually parting and joining all your channels that use that option, or just reconnecting 2015-06-20 21:45:31 dalias: /ns regain helps with that 2015-06-20 21:45:37 i couldn't get it to work 2015-06-20 21:45:41 still got the error 2015-06-20 21:45:45 interesting 2015-06-20 21:45:58 every time i used ns regain it gave me that error 2015-06-20 21:46:00 oh, they might be running an older version where it is still broken 2015-06-20 21:46:15 this will work, however 2015-06-20 21:46:26 /ns identify and then ghost 2015-06-20 21:46:45 ghost worked anyway 2015-06-20 21:46:48 actually it was already gone 2015-06-20 21:46:54 but /nick and /ns regain failed 2015-06-20 21:47:14 you can also auth to an account before changing your nick, using /ns identify [account] password 2015-06-20 21:47:21 which allows you to bypass that error 2015-06-20 21:47:34 since it's people doing +q $~a type bans 2015-06-20 21:49:27 i see 2015-06-20 21:49:37 for some reason my auto-ident script isn't working at all 2015-06-20 21:49:45 i should probably change it to use the long form with a 2015-06-20 21:49:49 i should probably change it to use the long form with account when i fix it tho 2015-06-20 21:49:59 possibly, or use SASL. 2015-06-20 21:50:22 :) 2015-06-20 21:50:23 it may be too new though, it is only 10 years old :P 2015-06-20 21:50:32 (in terms of IRC anyway.) 2015-06-20 21:51:10 :) 2015-06-20 21:51:33 does freenode still support iso646 national variants? ;-) 2015-06-20 23:12:11 wait 2015-06-20 23:12:17 busybox has an httpd 2015-06-20 23:12:49 busybox is really busy 2015-06-21 11:21:34 http://www.autistici.org/bakunin/codes.php , some nice old/new codes here 2015-06-21 11:23:02 any have used spread and pkg based on it, kinda looks ok, 2015-06-21 11:23:03 http://www.dsn.jhu.edu/software.html 2015-06-21 11:23:05 http://www.spread.org/apps.html 2015-06-21 14:04:53 hey guys, i have a huge problem. i build an raid 1 via mdadm 2015-06-21 14:05:04 it worked fine, but after a reboot it was gone 2015-06-21 14:05:11 why was that one not persistent?? 2015-06-21 14:05:33 and how can i get my raid back? 2015-06-21 14:10:36 hmm, i hope my data is not gone 2015-06-21 14:10:42 because that would be a problem 2015-06-21 14:21:39 leooooo: probably just not 'started' the raid 2015-06-21 14:21:44 can you do cat /proc/mdstat 2015-06-21 14:22:18 darkfaded: cat /proc/mdstat shows: Personalities: 2015-06-21 14:22:26 so it does not know any personalities 2015-06-21 14:22:39 modprobe md or what the name is 2015-06-21 14:22:46 maybe something is missing there 2015-06-21 14:23:01 anyway, don't worry, it's just not loaded and definitely not assembled / started 2015-06-21 14:23:18 dig for some howto (since i'm sleepy and a howto won't be sleepy) 2015-06-21 14:23:30 once you got everything active: 2015-06-21 14:23:34 during booting i get the follwoing error: udevd: IMPORT{buildin}: blkid unknown /lib/udev/rules.d/63-md-raid-array 2015-06-21 14:23:45 - make sure you have the raid drivers in your initrd 2015-06-21 14:23:52 - make sure your mdadm.conf is ok 2015-06-21 14:24:06 it should all go find after that 2015-06-21 14:24:24 my mdadm.conf is the default shipped one, i did not modify it 2015-06-21 14:25:00 you will have to. 2015-06-21 14:25:01 how do i check if my drivers are in the initrd? 2015-06-21 14:25:25 i don't know i didn't set up a sw raid today :) 2015-06-21 14:25:37 why do i have to modify it? md should assembly it based on the informations written to the superblocks 2015-06-21 14:26:09 ok then ask md 2015-06-21 14:26:50 autoassembly based on the metadata only is no longer supported because some dev thought it's not good 2015-06-21 14:27:10 go find and ask the md dev, it wasn't me 2015-06-21 14:27:14 ah, i see. so it's a alpine-linux special case? 2015-06-21 14:27:37 because in this file the first line is: "mdadm will function proper without this file 2015-06-21 14:29:29 i think it has something todo with: "udevd: IMPORT{buildin}: blkid unknown /lib/udev/rules.d/63-md-raid-array" 2015-06-22 08:00:16 morning 2015-06-22 08:01:07 And such a beautiful morning! 2015-06-22 08:01:31 it's raining cats and dogs here in the UK 2015-06-22 08:02:55 Good for the garden 2015-06-22 08:04:30 I watered that yesterday! Wish I hadn't bothered now :) 2015-06-22 08:14:13 haha indeed 2015-06-22 08:26:42 <_ikke_> It's nice weather here in the netherlands 2015-06-22 09:40:56 In Sweden today it was amazingly gorgeous 2015-06-22 09:41:21 And by today I mean the last three, since the sun hasn't set 2015-06-22 09:50:06 _ikke_: nice in the netherlands? you like cold and rain? 2015-06-22 09:51:39 <_ikke_> clandmeter: The sun is / was shining here 2015-06-22 09:52:50 you are probably not located in the south-east. 2015-06-22 09:52:58 <_ikke_> nope 2015-06-22 09:53:14 <_ikke_> It's getting more cloudy now though 2015-06-22 09:53:19 Seaside weather changes at the drop of a hat 2015-06-22 09:53:46 it should rain a bit though, gras is getting yellow here. 2015-06-22 09:55:14 Mostly I'm concerned with Amsterdam's cannabis crop 2015-06-22 09:56:23 haha 2015-06-22 09:56:57 ahills: most of it grows in my region ;-) 2015-06-22 09:57:14 ahills, it's not cultivated in the Netherlands nor in Sweden 2015-06-22 09:57:26 If it is, it's probably indoors 2015-06-22 09:57:49 Sweden I think doesn't have drugs, they can barely buy alcohol here 2015-06-22 09:58:05 It's even worse than the US which I didn't think was possible 2015-06-22 10:03:41 few months ago they found the largest chemical storage (non cleaning matterial) ever found in NL approx 100 meters from our office. it was included with a rocket launcher and some lighter devices.. 2015-06-22 10:04:12 haha 2015-06-22 10:04:20 A rocket launcher xD 2015-06-22 10:04:43 !!! 2015-06-22 10:05:00 Private space exploration! I hope they didn't confiscate 2015-06-22 10:06:13 kernel 4.1 seems interesting 2015-06-22 10:06:37 clandmeter, what's new? 2015-06-22 10:06:52 i never encrypted anything yet, does new ext4 enctyption support make it easier? 2015-06-22 10:08:01 i think it also includes some features to overlayfs 2015-06-22 10:13:43 Encryption is already very easy 2015-06-22 10:14:30 Likely experimental ext4 features will not be as nice as luks/cryptsetup 2015-06-22 11:05:35 Hello... I have a bit of a problem with my USB based Alpine Linux machine - in 'dmesg' it says that the USB media was not properly unmounted and that I should run 'fsck' - but I can't umount /media/usb (Resource busy). However, using 'lsof' I can't see anything actually using files from /media/usb ... 2015-06-22 11:05:40 Hello 2015-06-22 11:05:59 Any ideas what's using my /media/usb or how to find out? 2015-06-22 11:09:58 danci1973: modloop 2015-06-22 11:10:16 the modules are mounted 2015-06-22 11:10:33 i think you can stop modloop service and umount 2015-06-22 11:11:56 clandmeter: Thanks, will try right away... 2015-06-22 11:13:42 <_ikke_> otherwise, lsof / fuser 2015-06-22 11:13:53 clandmeter: It works... sort of. Stopping 'modloop' service umounts the USB drive, but now I run 'fsck.vfat /dev/usbdisk' and get 'Seek to 7863270912:Invalid argument'... 2015-06-22 11:15:07 never seen such msg 2015-06-22 11:15:31 clandmeter: in these troubled times you should be encrypting no matter what 2015-06-22 11:15:43 I guess I'll 'rsync' the content of the USB stick to my disk and then pull it out and check it on my machine... 2015-06-22 11:16:39 between hostile intelligence agencies and rising problems of hardware theft, the first thing i would do is encrypt any/all storage media ;) 2015-06-22 11:45:18 danci1973: fsck checks filesystem, not files 2015-06-22 11:48:23 hmm 2015-06-22 11:49:28 ncopa: is this somekind of bug? http://sprunge.us/RQIV 2015-06-22 11:50:26 suonds like apk upgrade/add went wrong 2015-06-22 11:50:39 try: apk fix 2015-06-22 11:50:46 or apk fix libflite 2015-06-22 11:50:50 its not installed 2015-06-22 11:51:06 was just checking lib and saw those 2015-06-22 11:51:15 'flite' 2015-06-22 11:51:21 is maybe the package 2015-06-22 11:51:32 yes we have a pkg like that 2015-06-22 11:51:40 it could be that ctrl-c was pressed while flite was installed 2015-06-22 11:51:51 ok 2015-06-22 11:51:59 apk add flite && apk del flite 2015-06-22 11:52:01 should fix it 2015-06-22 11:53:07 yeah, its gone 2015-06-22 11:53:44 so only after a correct apk add, the files get renamed? 2015-06-22 12:07:26 ahills: What do you mean? I am trying to check a filesystem... 2015-06-22 12:10:29 I was just making sure you weren't going to rsync to a new filesystem and fsck that one 2015-06-22 12:10:45 People with all levels of experience use IRC... 2015-06-22 12:11:02 ahills: Aaaahhh... I see. :) 2015-06-22 12:11:25 ahills: I was going to make a backup and re-create the FS, but... 2015-06-22 12:11:58 That's what I would do... 2015-06-22 12:12:09 I don't have time to debug ext4 2015-06-22 12:12:09 It seems that 'Seek to ....: Invalid argument' is down to the size of my USB stick - it's larger thatn 2GB and Alpine's version of fsck.vfat isn't quite ready for that... 2015-06-22 12:12:34 Oh, I think there is another FAT fsck 2015-06-22 12:12:41 fsck.msdos maybe? 2015-06-22 12:13:11 mkfs.msdos -F 32 or something like that is how I make those, iirc 2015-06-22 12:13:12 ahills: I tried all three that come with 'dosfstools' package... 2015-06-22 12:13:25 ahills: Oooo, -F 32 ? I missed that. 2015-06-22 12:13:30 You checked for extra flags, etc/ 2015-06-22 12:13:47 -F 32 to create a fat32 instead of fat16 2015-06-22 12:14:09 But I don't know if fsck has that 2015-06-22 12:14:33 I haven't ever run into this issue, fsck on alpine has always worked for me 2015-06-22 12:15:10 ahills: It doesn't look like fsck has an option for fat32 - I guess it *should* autodetect. 2015-06-22 12:15:25 I'm going to pull the stick now and see what my workstation says about it. 2015-06-22 12:15:35 Good luck! 2015-06-22 12:27:38 Interestingly, the latest OpenSuSE's fsck gives the same error... Stuck the stick in a Windows 7 laptop, it detected an error and I let it scan & fix... Now booting the Alpine machine. 2015-06-22 12:28:13 Yum, the error is gone... 2015-06-22 12:28:36 Isn't it scarry that Windows do something 'better' then Linux? :) 2015-06-22 12:40:53 danci1973: educate me. what does MS Windows do better? 2015-06-22 12:41:17 fscking their own filesystems 2015-06-22 12:41:32 Yup, that. :) 2015-06-22 12:46:02 :) is FAT really "theirs"? https://en.wikipedia.org/wiki/File_Allocation_Table 2015-06-22 12:50:36 ScrumpyJack: In order to overcome the volume size limit of FAT16, while at the same time allowing DOS real mode code to handle the format, Microsoft designed a new version of the file system, FAT32, which supported an increased number of possible clusters, but could reuse most of the existing code, so that the available conventional memory footprint was reduced by less than 5 KB under DOS. 2015-06-22 12:52:18 i think MS make *alot* money on android phones due to FAT 2015-06-22 13:04:48 <_ikke_> yup, at least from HTC 2015-06-22 13:04:53 <_ikke_> Perhaps others too 2015-06-22 13:05:15 <_ikke_> The joke was that MS earned more from android phones than from windows phones 2015-06-22 13:06:47 yes 2015-06-22 13:06:53 i think they do from all the big sellers 2015-06-22 13:06:56 samsung 2015-06-22 13:06:57 lg 2015-06-22 13:06:59 htc 2015-06-22 13:10:26 ncopa: could you rebuild taskd? it's broken at the moment 2015-06-22 13:12:41 pushed 2015-06-22 13:12:46 look for 1.1.0-r1 2015-06-22 13:16:03 thanks 2015-06-22 13:18:14 ERROR: taskd-1.1.0-r1: No error information 2015-06-22 13:19:09 apk update? 2015-06-22 13:19:16 naturally :) 2015-06-22 13:19:31 i'm using liskamm. it can see r1 2015-06-22 13:21:36 i ran rsync manually on the mirror and it's fine now 2015-06-22 13:21:51 funny that the APKINDEX showed r1 2015-06-22 13:21:59 (before the rsync) 2015-06-22 13:23:11 ncopa: could you rebuild libgnutls and libhogweed as well? taskd needs them (sorry, i didn't think) 2015-06-22 13:28:33 gnutls was already rebuilt 2015-06-22 13:28:38 what is the exact error message you get? 2015-06-22 13:29:44 hmm, what's the best way to upgrade one package? 2015-06-22 13:30:02 apk add -U -u 2015-06-22 13:33:36 after upgrade gnutls: http://sprunge.us/DcTG 2015-06-22 13:36:09 Nothing wrong with 'dosfstools' - somehow the partition on the USB drive was larger than the drive (fdisk reports 1487 cylinders, partition was 1505 cylinders). 2015-06-22 14:20:00 ScrumpyJack: looks like taskd itself 2015-06-22 14:22:53 ncopa: :( 2015-06-22 14:28:59 ScrumpyJack: after I did apk upgrade -U: 2015-06-22 14:29:02 ncopa-desktop:~$ taskd --version 2015-06-22 14:29:02 taskd 1.1.0 built for linux 2015-06-22 14:29:19 $ apk version taskd 2015-06-22 14:29:19 Installed: Available: 2015-06-22 14:29:19 taskd-1.1.0-r1 = 1.1.0-r1 2015-06-22 14:29:31 i cannot fix it more than that 2015-06-22 14:29:43 if you mix repos you will likely get problems 2015-06-22 14:30:11 libstdc++ and taskd versions needs to match 2015-06-22 14:30:21 ah yes, ok. 2015-06-22 14:34:05 right, it's not that I'm mixing repos, it's just that i'm resisting upgrading lately, so some packages are getting left behind. 2015-06-22 14:34:26 right, just upgraded 2015-06-22 14:34:33 i'll need to reboot i guess 2015-06-22 14:34:39 should not need reboot 2015-06-22 14:36:09 i have a new initramfs though 2015-06-22 14:39:32 my lxc containers don't start now :) 2015-06-22 14:40:42 failed to mount 'shm' on '/usr/lib/lxc/rootfs/dev/shm' 2015-06-22 14:40:55 thats a known issue 2015-06-22 14:41:25 lxc.mount.entry=shm dev/shm tmpfs nodev,nosuid,noexec,mode=1777,create=dir 0 0 2015-06-22 14:41:42 add that to your /var/lib/lxc//config 2015-06-22 14:42:08 it you have the shm dev/shm mount entry make sure that create=dir is there as mount option 2015-06-22 14:42:27 yeah, just saw that on github 2015-06-22 14:43:27 I just created a new lxc container & used the mount entries from that 2015-06-22 16:06:30 ncopa: do you have any plans on actually providing static lib archives .a in like additional packages, say -staticlibs? 2015-06-22 16:06:44 ncopa: so far I see you guys just strip them. :< 2015-06-22 16:34:33 I always leave them in -dev 2015-06-22 16:39:21 that could work, too 2015-06-22 16:39:28 iirc openssl at least does not have static archive 2015-06-23 06:22:22 slashbeast: we normally just purge the static libs 2015-06-23 06:22:43 we could probably provide static libs for specific projects if needed but in general we try avoid static linking 2015-06-23 06:23:02 i suppose a -static subpackage is an option 2015-06-23 06:26:31 how hard is it to build a package 2015-06-23 06:26:40 if it' not too painful I've half a mind to build ag 2015-06-23 06:29:43 ncopa: well, if that woudn't cost you too much effort, then it would be great 2015-06-23 06:30:40 ncopa: I use alpine to link statically a lot of software, as I do have some deployments in company that simply does not have the needed libs, and shipping whole rootfs is maddness 2015-06-23 06:31:01 ncopa: if thats however too much for so little gain, then thats fine too, I could try to rebuild needed packages. 2015-06-23 06:33:03 slashbeast: do you have a list of relevant packages? 2015-06-23 06:33:13 i doubt you need it for all 1700 aports 2015-06-23 06:33:51 i'll look at making it simple as adding $pkname-static to subpackages 2015-06-23 06:34:31 ncopa: no list so far, I will try to get it done and sent it to you 2015-06-23 06:36:21 feel free to ssend it privately if you want 2015-06-23 07:23:39 Hello... Is there a way to influence how network interface names correspond to actual hardware / ports? For exampel - I have a motherboard with 4 onboard NICs, which used to be eth0-eth3. Now I added a couple of PCI cards and now these are eth0-eth3, while onboard NICs got eth4-eth7 ... 2015-06-23 07:31:54 you could use udev rules for that 2015-06-23 07:36:53 ScrumpyJack: Yeah, already looking into that... 2015-06-23 07:37:07 danci1973: if you use mdev you can use iproute2 (even the busybox one) to rename interfaces 2015-06-23 07:37:12 or use mactab 2015-06-23 07:37:34 slashbeast: mactab? 2015-06-23 07:37:41 slashbeast: Never heard of that... 2015-06-23 07:39:06 I like 'mactab'. :) 2015-06-23 07:39:06 danci1973: check mdev-like-a-boss 2015-06-23 07:39:18 there's a 'settle-nics' scripts that I've wrote for that solo purpose 2015-06-23 07:39:43 https://github.com/slashbeast/mdev-like-a-boss/blob/master/helpers/settle-nics 2015-06-23 07:40:18 i really need to learn to use mdev and/or udev well... 2015-06-23 07:41:07 btw is there a way to stick to mdev and avoid udev when X stuff wants/tries to pull in udev? 2015-06-23 07:41:21 it seems like a nasty situation to be in, where adding packages could break your config.... 2015-06-23 07:41:49 well, I do run gentoo, with X and I do use mdev 2015-06-23 07:41:57 if you disable hotplugging, you can use X without udev 2015-06-23 07:42:04 Is 'mdev' a package in Alpine or is it 'inherent'? 2015-06-23 07:42:07 I think in Alpine you could just install but not use udev 2015-06-23 07:42:17 mdev is a package 2015-06-23 07:42:38 https://github.com/slashbeast/mdev-like-a-boss thats what I use on my work laptop as well as private one 2015-06-23 07:42:41 not a single issue 2015-06-23 07:43:04 fwiw libudev is needed for Mesa, at least for linking lol 2015-06-23 07:44:07 slashbeast, hotplugging what? 2015-06-23 07:44:14 keyboards? 2015-06-23 07:44:47 works flawless, and mouse goes thru mouse driver /dev/input/mice 2015-06-23 07:44:49 I'm building a router/firewall, so my biggest concern are NICs. 2015-06-23 07:45:10 https://github.com/slashbeast/mdev-like-a-boss/blob/master/xorg.conf.d/30-mouse.conf 2015-06-23 07:45:15 touchpad with synaptics also works just fine 2015-06-23 07:45:23 slashbeast, yeah i figured if you just use /dev/input/mice there would be no need for explicit hotplug that the X server is aware of 2015-06-23 07:45:47 dalias: indeed. I hope that libevent at some point will be able to hotplug without udev 2015-06-23 07:46:04 I love simple solutions. I had huge problems with udev to be honest 2015-06-23 07:46:59 dalias: apart from some virtualbox rules that of course needs to be ported to mdev, I did not had any real issue for using mdev 2015-06-23 07:47:57 life was better before udev/mdev/dbus/networkmanagers/and dependancy hacks for init 2015-06-23 07:49:03 the only thing I run with dbus is wicd (wireless) 2015-06-23 07:49:08 and openrc as init 2015-06-23 07:49:14 there's not much things that can broke here you know 2015-06-23 08:00:37 slashbeast: settle-nic - do I just add it to default runlevel or should it be included in a more specific place? 2015-06-23 08:12:17 danci1973: I run it via mdev, check mdev.conf that I've provided in the said repo 2015-06-23 09:11:28 slashbeast: Thanks, got it working. 2015-06-23 09:11:36 Have I said I like '/etc/mactab'? :D 2015-06-23 09:11:58 yup 2015-06-23 09:12:15 in udev you have 70-something-net.rules in /etc/udev for the same thing 2015-06-23 09:22:02 Yes, I know udev, but I'd like to keep the firewall/router Alpine down to a minimum... 2015-06-23 09:29:00 ++ 2015-06-23 09:40:41 what's the matter with the order of the NICs? 2015-06-23 09:41:40 one want to have predictable nic names. 2015-06-23 09:42:57 and not boot a server with anotehr NIC to just find that the one that used to be WAN is now the new nic 2015-06-23 13:15:37 anyone using wicd 2015-06-23 14:07:36 yes, i think slashbeast uses it 2015-06-23 14:12:04 i use wicd-curses on debian on a powerpc 2015-06-23 14:12:39 but other than that i just use wpa_supplicant 2015-06-23 14:12:50 mhe: what gives? 2015-06-23 14:17:25 mhe http://it-offshore.co.uk/linux/alpine-linux/46-alpine-linux-automatic-wireless-configuration 2015-06-23 14:19:37 yeah, that's what i do. one script per wifi access point i connect to. can't be bothered to make a menu 2015-06-23 15:04:37 Trying to install redmine, I found a bug in the default configuration file provided for unicorn. The group defined in redmine is "nogroup" but "redmine" is declared in the /etc/unicorn/redmine.conf.rb 2015-06-23 15:05:43 therefore, unicorn fails to start. 2015-06-23 15:06:09 It is easy to fix (just changing redmine to no group seems to do the trick) 2015-06-23 15:17:00 Jean-Scotch: what do you mean "in redmine"? 2015-06-23 15:17:42 I mean in /etc/passwd, the redmine user is "redmine:nogroup" 2015-06-23 15:17:50 sorry, was not precise enough 2015-06-23 15:18:24 in /etc/unicorn/redmine.conf.rb, it is defined as redmine:redmine 2015-06-23 15:18:43 ah ok 2015-06-23 15:18:54 unicorn complains at startup that the group does not exist 2015-06-23 15:19:50 http://git.alpinelinux.org/cgit/aports/tree/main/redmine/redmine.pre-install 2015-06-23 15:21:52 maybe something goes wrong with that pre-install? 2015-06-23 15:25:11 I suppose so 2015-06-23 15:26:29 I guess the group should be added before the user... 2015-06-23 15:28:18 it does 2015-06-23 15:28:46 the last addgroup is to add the user to the group 2015-06-23 16:04:46 the user redmine is added to the group www-data. But nowhere the primary group of user redmine is defined as "redmine". Or maybe I read wrong... 2015-06-23 16:06:58 I have to go. See you... 2015-06-24 06:59:17 morning campers 2015-06-24 07:15:47 what's a good way of getting time on a rpi? 2015-06-24 07:16:40 ScrumpyJack, what time? 2015-06-24 07:16:45 system time? 2015-06-24 07:16:49 uptime? 2015-06-24 07:16:54 ntp time? 2015-06-24 07:17:00 cpu usedtime? 2015-06-24 07:17:01 ;) 2015-06-24 07:18:03 by that i mean, the rpi has no rtc, of course, and some ntp implementations struggle (or default are configured to "panic") when setting a new system time offset by 45 years :) 2015-06-24 07:18:23 coffee time 2015-06-24 07:27:13 back 2015-06-24 10:49:06 do we have any bacula gui in the repos? 2015-06-24 11:03:16 hey guys, is it possible to use the bash-completions also with the ash shell? 2015-06-24 11:27:52 leo-unglaub, http://www.linfo.org/command_completion.html 2015-06-24 11:28:07 read the last paragraph 2015-06-24 11:28:39 been trying the fish shell. boy, what a pain that shell is 2015-06-24 11:28:57 ScrumpyJack, not a good shell? 2015-06-24 11:29:26 alacerda: thanks! 2015-06-24 11:29:32 yw 2015-06-24 11:30:02 zsh is rather powerful, but require you to understand few things that you can no longer just echo =foo as = have special meaning 2015-06-24 11:30:09 i suppose it would take getting used to, but for me, it's too noisy. too much feedback gets in the way 2015-06-24 11:31:24 slashbeast: yeah, i love zsh. it's replaced tcsh for me, since i failed to build tcsh on alpine linux and alpine linux is now my env of choice :) 2015-06-24 11:33:29 for you bash users, what do you export as TERM? 2015-06-24 11:33:38 i have problems with colored output on the alpine bash shell 2015-06-24 11:34:11 my stuff looks like [[33mcommit 5f01bd72360c327f75272973eaa5a97907e5a1b2[[m 2015-06-24 11:34:52 sounds like terminfo issue 2015-06-24 11:39:03 speaking on which ... check this: http://bashrcgenerator.com/ 2015-06-24 11:43:06 slashbeast: yeah, do you have any idea on how to fix it? 2015-06-24 11:44:52 is the $TERM correct and do you have /usr/share/terminfo for specified $TERM? 2015-06-24 11:46:12 leo-unglaub: apk add less 2015-06-24 11:46:29 less is more :) 2015-06-24 11:48:00 clandmeter: haha, perfect :) 2015-06-24 11:48:15 i ma going to put this in the wiki, because i am sure i will have the question in 5 month again *g* 2015-06-24 16:51:28 i have added chronyd and tor to the default runlevel, but both dont start on boot 2015-06-24 16:51:46 is there something else i have to do beside movint it to the correct runlevel 2015-06-24 16:54:01 no, but if you're using lbu, make sure to lbu commit 2015-06-24 16:54:17 ahills: i have alpine installed on my hdd 2015-06-24 16:54:27 so i dont need that, right? 2015-06-24 16:54:56 leo-unglaub: do they show in rc-update output? 2015-06-24 16:55:17 darkfaded: yes, they do. both in "default" 2015-06-24 16:56:29 i have the same problem on two devices 2015-06-24 16:56:40 so either i fucked something up or there is a bug 2015-06-24 16:58:40 bug prolly :) 2015-06-24 16:59:24 do i have to regenerate initrams in order to get it working? 2015-06-24 16:59:44 because i have three devices in my crypttab but only the first one gets luksOpened on boot 2015-06-24 16:59:50 the other ones are simply not tryed 2015-06-24 17:00:04 so i think somehow alpine is using "old" from init information 2015-06-24 17:00:41 i think both issues are related 2015-06-24 17:00:43 you have to add dmcrypt or something to the boot runlevel 2015-06-24 17:00:58 apk info -L cryptsetup 2015-06-24 17:00:59 ahills: it works for the first device, but not the other two 2015-06-24 17:01:13 it simply does not attempt do do it because it must use wong infirmations 2015-06-24 17:01:17 well it should work for your root device because of your kernel command line 2015-06-24 17:01:29 thats probobly the same reason why tor and chronyd dont start at the default runlevel 2015-06-24 17:07:42 re 2015-06-24 22:44:20 in addition how are people storing standard unit values, i.e. ms, kg etc along with the value. Is it being stored as a tag 2015-06-24 23:02:03 jalatera_: ? 2015-06-24 23:33:19 clandmeter: wrong group 2015-06-24 23:38:09 how do i set the timezone on an alphine distribution? 2015-06-24 23:39:12 does setup-timezone not work 2015-06-24 23:50:51 exit 2015-06-24 23:55:14 ammunta: this is how i am calling it setup-timezone -z Australia/Melbourne 2015-06-25 00:03:22 what goes wrong? 2015-06-25 07:52:22 morning 2015-06-25 07:52:50 annoyingly, i seem to have lost scrollback in my X terminals 2015-06-25 07:53:12 scrolling up with the mouse only shows a few lines of text 2015-06-25 07:54:37 regardless of terminal emulator 2015-06-25 08:04:07 regardless of window manager 2015-06-25 08:05:09 ScrumpyJack: disk full? :)))))\ 2015-06-25 08:19:09 it looks like it's a display manager issue. if i run startx with my display manager, i'm fine 2015-06-25 08:19:19 the display manager is slim 2015-06-25 08:19:23 wierd 2015-06-25 08:22:46 different issue. what comes first in openrc. lvm or mounting stuff? 2015-06-25 08:23:35 maybe i should put lvm in boot runlevel 2015-06-25 12:35:46 rnalrd: hey ho. would you mind awefully creating an owncloud-cmd subpackage of the owncloud-client package? 2015-06-25 12:36:29 so that someone could use owncloudcmd (command line client) without the need for qt! :) 2015-06-25 13:49:14 Any Ideas why I'm getting an apk conflict between freetype and freetype-inifinality ? 2015-06-25 13:50:50 freetype-infinality-2.5.5-r1 provides: 2015-06-25 13:50:50 so:libfreetype.so.6=6.11.4 2015-06-25 13:50:56 freetype-2.6-r1 provides: 2015-06-25 13:50:57 so:libfreetype.so.6=6.12.0 2015-06-25 13:51:22 Orochimarufan, if both packages provide same name with version, there's automatic conflict between them 2015-06-25 13:51:38 so there's nothing I can do? 2015-06-25 13:51:56 sounds like freetype-infinality is freetype with a patch applied 2015-06-25 13:52:09 also, it complains about freetype-2.5.5-r0 and freetype-infinality-2.5.5-r1 2015-06-25 13:52:11 it is 2015-06-25 13:52:14 freetype-infinality should probably provide freetype too 2015-06-25 13:52:27 so any packages depending on freetype explicitly would be satisfied 2015-06-25 13:52:34 sounds like packaging error 2015-06-25 13:52:48 freetype-infinality-2.5.5-r1: 2015-06-25 13:52:48 conflicts: freetype-2.5.5-r0[so:libfreetype.so.6=6.11.4] 2015-06-25 13:53:43 something tries to pull freetype 2015-06-25 13:53:53 that conflict somehow prevents me to build some packages 2015-06-25 13:54:01 so if freetype-infanality would provide freetype, it'd be ok 2015-06-25 13:54:16 it's more likely something tries to pull infinality, since I already have the normal one installed 2015-06-25 13:54:43 true 2015-06-25 13:54:47 freetype policy: 2015-06-25 13:54:47 2.5.5-r0: 2015-06-25 13:54:47 lib/apk/db/installed 2015-06-25 13:55:06 what does the error report say. what freetype-infinality satiesfies? 2015-06-25 13:55:10 maybe it's really about the 2.6 in the repos 2015-06-25 13:55:30 2.6-r1: 2015-06-25 13:55:30 http://nl.alpinelinux.org/alpine/edge/main 2015-06-25 13:55:47 possible. maybe you are mixing and matching different repos? 2015-06-25 13:55:54 the satisfies are the same for both, except for freetype-dev vs freetype-infinality-dev 2015-06-25 13:56:12 nope, just edge/{main,testing} 2015-06-25 13:56:34 (plus custom-built packages, but those shouldn't matter, freetype isn't one of them) 2015-06-25 14:04:14 ACTION is retarded 2015-06-25 14:04:43 actually upgrading it fixed the issue 2015-06-25 14:29:48 also ncopa, any news on the per-arch source arrays? 2015-06-25 14:30:28 Orochimarufan: no, i have not had time to look at it 2015-06-25 14:30:49 It's all right, I'll just continue with my hacks ;) 2015-06-25 14:32:27 Hello. I am having a problem with the apk program. I installed on Raspberry Pi SD Card. Worked great. I looked at the files on the respository using web browser. Many of the files when subject to "apk add " generate an error about unsatisfiable constraint. Had about 30% success. What am I doing wrong? 2015-06-25 14:33:01 ASTi: did you run "apk update && apk upgrade"? 2015-06-25 14:33:12 I've run into that same issue a few minutes ago 2015-06-25 14:34:10 Yes, I did. As an example of one that doesn't work: clang. 2015-06-25 14:34:24 what exactly does it complain about? 2015-06-25 14:35:10 ERROR: unsatisfiable constraints:\n\tclang (missing):\n\t\trequired by world[clang] 2015-06-25 14:36:05 are you sure clang is in one of your repos? 2015-06-25 14:36:08 It gives this same error for many of the packages. I tried downloading the .apk locally ad "apk add " and it worked, but it couldn't resolve the dependencies locally. 2015-06-25 14:36:13 ASTi, http://pkgs.alpinelinux.org/packages?package=clang&repo=all&arch=armhf 2015-06-25 14:36:40 http://dl-2.alpinelinux.org/alpine/v3.1/main/armhf/clang-3.5.0-r0.apk 2015-06-25 14:37:11 cat /etc/apk/repositories 2015-06-25 14:37:18 are you using this repo? 2015-06-25 14:37:56 No. Using the dl-2 one. I'll switch it: thanks. 2015-06-25 14:38:08 There is no clang in current aports :O 2015-06-25 14:38:36 not even in unmaintained 2015-06-25 14:39:08 actually, it's a subpackage of llvm 2015-06-25 14:39:50 How come it is on this page then? http://dl-2.alpinelinux.org/alpine/v3.1/main/armhf ? 2015-06-25 14:40:08 ASTi, you can wget the package and apk add 2015-06-25 14:40:32 As I said, It's a subpackage of llvm, so I couldn't find it at first 2015-06-25 14:40:47 Yes, I was able to do this, but you have to do the dependencies manually. 2015-06-25 14:41:03 ASTi, are you using AL 3.1 ? 2015-06-25 14:41:20 And according to http://git.alpinelinux.org/cgit/aports/tree/main/llvm/APKBUILD it should be built on armhf just fine :O 2015-06-25 14:41:24 what is you /etc/apk/repositories content? 2015-06-25 14:42:09 It was mesa I was trying to get which led to llvm which led to clang. apk couldn't do any of them. uname -r returns "3.18.12-0-rpi2" 2015-06-25 14:42:45 what about cat /etc/os-release ? 2015-06-25 14:44:06 Doh! I have http://dl-2.alpinelinux.org/alpine/v3.2/main. Should have v3.1. My bad. I'll switch it and retry. Thanks for the help. Will come back if it doesn't work. Otherwise: operator error. Thanks again! 2015-06-25 14:44:14 ;) 2015-06-25 14:45:50 Oh, perhaps too soon. /etc/os-release says VERSION_ID=3.2.0, PRETTY_NAME="Alpine Linux v3.2", hmmm..... 2015-06-25 14:46:46 So on the v3.2 page, there is no clang. On the v3.1 page, there is clang. Can I use the v3.1 stuff with a release that says 3.18 from uname and 3.2 from os-release? 2015-06-25 14:53:52 you can certainly try :) 2015-06-25 15:18:13 can I use multiple --repository arguments when doing apk.static/chroot install? 2015-06-25 15:22:49 it appears so :-) 2015-06-25 15:31:50 is it safe to use edge/testing packages on a stable release 2015-06-25 15:34:36 pin edge http://wiki.alpinelinux.org/wiki/Alpine_Linux_package_management#Repository_pinning 2015-06-25 15:40:46 Ok, so I got past the earlier issue by using the 3.1 repository. Stuff seemed to install ok. Thanks to all. 2015-06-25 16:01:17 wait, are there https repos? 2015-06-25 16:03:12 <_ikke_> I'm not aware of any, but that wouldn't add a lot, because all packages and indexes are signed 2015-06-25 16:06:27 actually are there even ftp repos? 2015-06-25 16:09:35 <_ikke_> perhaps one of the mirrors offers ftp 2015-06-26 07:01:37 ted about? 2015-06-26 07:11:45 ASTi: clang/llvm is currently broken on arm 2015-06-26 07:12:00 #4372 2015-06-26 07:12:23 i thought i had a fix/workaround for it yesterday, but no... :-/ 2015-06-26 08:30:05 we should probably get a package for this https://github.com/xanmanning/alarm-fake-hwclock 2015-06-26 08:30:22 as there's no real way to save the clock on a RP if you're missing a RTC clock 2015-06-26 08:32:33 why not just disable hwclock altogether then? 2015-06-26 08:35:22 oh i do actually need my clock to not be 1970 2015-06-26 08:35:35 because it causes problems with the VPN coming up if the time is older than certificate creation 2015-06-26 08:36:16 so saving whatever the time "was" when it shut down and setting it to "that" when it starts up and then letting ntpd handle it more precisely once the system is online seems like the best solution 2015-06-26 08:36:59 i'd sort of written a crude script to do it, but using that is probably a better idea because it's designed specifically for that purpose 2015-06-26 08:37:47 <_ikke_> sillysausage: Yeah, i've done something similar 2015-06-26 08:37:53 <_ikke_> crude script 2015-06-26 08:38:16 unfortunately i don't have an alpinelinux dev environment atm 2015-06-26 08:38:25 but the above thing seems like what we want 2015-06-26 08:38:49 also seems to be what both debian and arch are using 2015-06-26 08:40:46 i wonder how easy it would be to port the sysvinit script to openrc 2015-06-26 08:41:22 https://dpaste.de/TbEo/raw 2015-06-26 08:43:55 might just follow this http://big-elephants.com/2013-01/writing-your-own-init-scripts/ 2015-06-26 08:44:03 and test it locally and if it works someone else can package it 2015-06-26 08:44:25 should be pretty trivial to package, as it's only one C file 2015-06-26 08:44:26 <_ikke_> I can try to make a package later if you have the rc 2015-06-26 08:44:31 yeah 2015-06-26 08:44:53 <_ikke_> I have a bit of time in about 2 hours 2015-06-26 08:55:46 ill get started on it then 2015-06-26 09:11:10 sillysausage: some ntp software support adjusting the clock when it starts up even if the offset is silly 2015-06-26 09:11:32 yeah i've just got ntpd 2015-06-26 09:11:32 or are you trying to adjust the time before you get a network connection? 2015-06-26 09:11:38 correct 2015-06-26 09:11:46 and that wont also be possible because the VPN is not yet up 2015-06-26 09:12:02 or PPP session 2015-06-26 09:12:27 ideally we just save a time when when we shut down that is newer than the creation time of the certificates 2015-06-26 09:12:43 so you're using a rasbpi w/ Alpine Linux as a VPN endpoint? 2015-06-26 09:13:01 main reason i went away from doing my hackjob script to this is because i had been saving a time (but not changing it on shutdown) 2015-06-26 09:13:13 then my dnscrypt server changed it's certificates 2015-06-26 09:13:45 so setting it later than the VPN certs was now no longer good enough because that was before the dnscrypt certificates were created 2015-06-26 09:13:57 so yeah ideally just shutting down and saving a current time is the best way 2015-06-26 09:14:02 ScrumpyJack: yes. 2015-06-26 09:14:18 that way it can work irrespective of whether dnscrypt or the vpn certificates have changed 2015-06-26 09:14:31 because it's unlikely they will change in the time i shutdown and start up 2015-06-26 09:15:24 i'd use one of these http://tinyurl.com/naqmrax 2015-06-26 09:16:45 oh sure, i could buy an RTC clock. 2015-06-26 09:16:54 but i would have to wait for it to get here. 2015-06-26 09:20:19 there is a cool GPS module as well. I built a stratum 1 ntp server with a rasbpi here for a LAN. no network needed, time straight from satellite's atomic clock :) 2015-06-26 09:21:02 heh 2015-06-26 09:21:08 yeah but that kind of sucks if it's indoors ;) 2015-06-26 09:21:46 GPS antenna on the roof! :) 2015-06-26 09:25:40 works nicely too :) 2015-06-26 10:25:40 <_ikke_> sillysausage: I have a bit of time now to look into it 2015-06-26 10:29:19 yep i've been writing the init script 2015-06-26 10:29:41 https://github.com/xanmanning/alarm-fake-hwclock 2015-06-26 10:29:57 should be simply make/make install 2015-06-26 10:30:22 pop the binary in /usr/sbin and the man file in docs 2015-06-26 10:30:36 or split it into two packages eg a -docs one 2015-06-26 10:30:43 <_ikke_> yeah 2015-06-26 10:30:44 i think that's common with alpinelinux isn't it 2015-06-26 10:30:48 <_ikke_> yup 2015-06-26 10:31:08 the rc script is a oneshot kine of script 2015-06-26 10:31:13 as it's not a daemon or anything 2015-06-26 10:31:54 my first openrc script btw https://dpaste.de/g18F/raw 2015-06-26 10:32:11 im in the process of testing it now 2015-06-26 10:33:42 <_ikke_> is this only arm? 2015-06-26 10:33:43 i did see this wrapper in debian 2015-06-26 10:33:48 can be either 2015-06-26 10:33:50 <_ikke_> ok 2015-06-26 10:33:57 but i would think you would probably only want it on arm 2015-06-26 10:34:05 i guess there might be x86 arches that don't have rtc clocks 2015-06-26 10:34:13 so no reason to limit it 2015-06-26 10:34:19 (does compile on both) 2015-06-26 10:34:42 debian had this wrapper script 2015-06-26 10:34:47 https://dpaste.de/q9yx 2015-06-26 10:35:26 im still testing it, when i used it before it worked 2015-06-26 10:36:54 <_ikke_> What version? There are two tags 2015-06-26 10:37:39 <_ikke_> 0.1 and 0.1b 2015-06-26 10:37:39 oh it looks like debian has expanded it 2015-06-26 10:37:46 i think use debian's version 2015-06-26 10:37:52 they've expanded the man file too 2015-06-26 10:38:04 <_ikke_> ah, you mean, with git describe? 2015-06-26 10:38:19 <_ikke_> Or you mean, expanded functionality 2015-06-26 10:38:27 i mean it seems to work properly :P 2015-06-26 10:38:29 lol 2015-06-26 10:38:38 https://packages.debian.org/unstable/main/fake-hwclock 2015-06-26 10:38:44 as in has that wrapper script and works 2015-06-26 10:38:49 where as when i compiled the thing from git 2015-06-26 10:38:52 it didn't seem to work 2015-06-26 10:39:12 http://http.debian.net/debian/pool/main/f/fake-hwclock/fake-hwclock_0.9.tar.gz 2015-06-26 10:39:16 is where i grabbed it from 2015-06-26 10:39:28 <_ikke_> is it the same project though? 2015-06-26 10:39:43 <_ikke_> that github page says it was based on fake-hwclock 2015-06-26 10:39:47 http://git.einval.com/cgi-bin/gitweb.cgi?p=fake-hwclock.git 2015-06-26 10:39:52 oh 2015-06-26 10:39:55 hmm 2015-06-26 10:39:59 let me check 2015-06-26 10:40:19 oh shit yeah you're right 2015-06-26 10:40:22 that's alarm-fake-hwclock 2015-06-26 10:40:32 i totally just did not see the "alarm" there. 2015-06-26 10:40:41 yeah we want the debian one, im not sure what that one does exactly 2015-06-26 10:41:05 <_ikke_> Ok, so you want fake-hwclock? 2015-06-26 10:41:19 yeah 2015-06-26 10:41:25 i found the debian one with their shell script works 2015-06-26 10:41:48 for some reason i couldnt' get the one from xan.manning to work 2015-06-26 10:42:17 looks like it's a simple shell script 2015-06-26 10:42:56 yeah im not really sure what that C implementation does, it doesn't seem to work though 2015-06-26 10:43:27 <_ikke_> So no need to build anything 2015-06-26 10:43:35 correct. 2015-06-26 10:44:03 just pop fake-hwclock aka https://dpaste.de/q9yx into a package 2015-06-26 10:44:15 the openrc script i wrote will still work 2015-06-26 10:44:28 https://dpaste.de/g18F/raw 2015-06-26 10:44:49 http://git.einval.com/cgi-bin/gitweb.cgi?p=fake-hwclock.git;a=tree all of it can just be taken out of git 2015-06-26 10:46:55 i shall upload this to my router 2015-06-26 10:55:07 <_ikke_> hmm, weird issue. Getting cannot cd to dir 2015-06-26 10:55:30 <_ikke_> in the prepare step 2015-06-26 10:56:06 have you set $_builddir ? 2015-06-26 10:56:13 in the APKBUILD file? 2015-06-26 10:56:27 <_ikke_> yes 2015-06-26 10:56:50 <_ikke_> the tar contains a dir called git 2015-06-26 10:57:00 <_ikke_> so I have $_builddir=git 2015-06-26 10:57:35 <_ikke_> /usr/bin/abuild: cd: line 1: can't cd to git 2015-06-26 10:57:42 can i see your APKBUILD file? 2015-06-26 10:58:29 <_ikke_> http://sprunge.us/Xdec 2015-06-26 10:59:15 abuild unpacks to $srcdir 2015-06-26 10:59:56 so you need _builddir=${srcdir}/somethin 2015-06-26 11:00:55 <_ikke_> ah ok. Tried src/git, but that didn't work. Needs to be a absolute path I guess 2015-06-26 11:01:43 use _builddir=${srcdir} literally 2015-06-26 11:01:56 abuild will replace the $srcdir 2015-06-26 11:01:58 <_ikke_> yeah, used that 2015-06-26 11:02:30 generally, you would use _builddir=${srcdir}/${pkgname}-${pkgver} 2015-06-26 11:03:19 <_ikke_> right, but that requires that name to be in the tarbal, right? 2015-06-26 11:05:01 you can ask abuild to save-as a specific name 2015-06-26 11:06:14 ah i see 2015-06-26 11:06:16 hang on 2015-06-26 11:06:55 _builddir=$srcdir/git worked for me 2015-06-26 11:16:19 <_ikke_> yeah, used that and it works now 2015-06-26 11:17:42 hey, i have a problem with time. i uses setup-timezone to set my timezone to Europe/Vienna but the system time is still Fri Jun 26 10:56:29 CEST 2015 instead of 13:17 2015-06-26 11:17:59 i have restarted chronyd twice 2015-06-26 11:18:08 and it updates the time, but always to a wron value 2015-06-26 11:47:06 i don't know much about chrony, but it should have an interface to it's peers 2015-06-26 11:47:15 can you check what peers are saying? 2015-06-26 11:48:38 chronyc looks like the cmdline query proggy 2015-06-26 11:49:31 ScrumpyJack: is the system time synced to the hardware clock on shutdown? 2015-06-26 11:49:35 it should, right? 2015-06-26 11:50:13 that would be down to shutdown scripts i guess 2015-06-26 11:51:10 leo-unglaub: you can also try rdate 2015-06-26 11:52:48 if chrony behaves anything like standard ntp, then it will update the time in steps if the offset is over a certain value 2015-06-26 11:53:26 perhaps that's what's happening. you can override this of course 2015-06-26 11:53:30 clandmeter: nice tip, thanks! 2015-06-26 11:55:28 if you are using ntp, it's best to diable hwclock 2015-06-26 11:55:56 so not let the hardware clock be set at shutdown 2015-06-26 11:58:07 i looks like chronyd supports iburst, so add that to your server config line 2015-06-26 12:04:49 chrony look pretty neat actually 2015-06-26 12:10:55 ScrumpyJack: yeah, its very small and nice 2015-06-26 12:11:07 i just needes something easy to keep my tor relay in time sync 2015-06-26 12:11:24 tor on Alpine Linux? 2015-06-26 12:11:31 (I need to make one) 2015-06-26 12:11:44 ScrumpyJack: yes, all my servers run alpine 2015-06-26 12:11:50 company and private 2015-06-26 12:12:00 i'm getting there too 2015-06-26 12:12:11 alpine is the perfect distro for it 2015-06-26 12:12:20 no bload and relativly secure 2015-06-26 12:12:51 put rtcsync 2015-06-26 12:13:16 in your chrony conf file if you want your hardware clock set by ntp 2015-06-26 12:15:02 i like the server moo "offline" directive in chrony. neat 2015-06-26 12:16:33 ScrumpyJack: thanks for rtcsync 2015-06-26 12:18:01 yep, chrony slows down or speeds up the clock as some software (especially mail servers) doesn't like big time jumps 2015-06-26 12:18:41 so my procedure on a fresh system is to set the time initially with "ntpdate pool.ntp.org" and then let chrony keep it accurate 2015-06-26 12:20:00 jomat: good tactic! 2015-06-26 12:40:13 ncopa: you have packages iptables, i have now packages the successor of it, nftables. can i send you the package for submission? 2015-06-26 12:40:34 sure 2015-06-26 12:40:53 git send-email to alpine-aports@lists.alpinelinux.org 2015-06-26 12:41:18 alright, will do 2015-06-26 12:41:56 i have also updates the mariadb and the hiawatha package because i uses the on a daily base in my company, can i submit the patches there as well 2015-06-26 12:42:16 or can i take ownership of them if no one else wants to? 2015-06-26 12:48:55 please keep sending patches for now and ask again for maintainership in 2 months 2015-06-26 12:49:06 i'd love to give you maintainership 2015-06-26 12:49:27 i just dont want hand it over to someone who sticks around for 1 month and then leave 2015-06-26 12:50:08 leo-unglaub: please feel free to join #alpine-devel too 2015-06-26 13:24:20 ncopa: yeah, thats absolutly correct :) 2015-06-26 13:24:38 it took me 2 years to get maintainership on my old distro 2015-06-26 13:24:46 but then i did it for 6 years :) 2015-06-26 13:24:57 i just send you patches and maybe in half a year we talk again 2015-06-26 13:27:05 hopefully i see more patches before 6 months :) 2015-06-26 13:28:47 ncopa: thats the list of packages i have patched up locally and maintain them for the use on our company servers: http://dpaste.com/0XZA0WP.txt 2015-06-26 13:29:17 if you are interrested in getting missing packages into alpine main from that list, i send you the patches tonight 2015-06-26 13:29:33 i am 2015-06-26 13:30:40 actually, thunderbird is missing on my list .. but i am trying to get off thunderbird so i did not add it to the list 2015-06-26 14:28:24 hmm do you have a link to my dpaste earlier 2015-06-26 14:28:34 i accidentally lost the openrc script :( 2015-06-26 14:29:12 would be nice if someone who was here a few hours ago could re-paste it for me :) 2015-06-26 14:29:30 _ikke_: how were you going with the APK? 2015-06-26 14:32:44 oh we might have something already /etc/init.d/swclock 2015-06-26 14:33:14 maybe we should have been using that _ikke_ instead 2015-06-26 14:33:21 it seems it's a part of openrc 2015-06-26 14:34:36 https://wiki.gentoo.org/wiki/OpenRC/Baselayout_1_to_2_migration#Clock 2015-06-26 14:34:46 ill test that might not be any reason to package fake-hwclock then 2015-06-26 14:34:52 if that can do it and is a part of openrc 2015-06-26 14:35:25 ScrumpyJack: was "ted about?" directed to me? 2015-06-26 14:39:51 although it depends on where swclock saves the time i guess 2015-06-26 14:40:49 <_ikke_> sillysausage: If you provide the rc, I have a packages. But let me know if it's still needed 2015-06-26 14:41:03 yeah i posted it here a few hours ago 2015-06-26 14:41:07 in a dpaste.de link 2015-06-26 14:41:17 but i err lost it when i rebooted my router, i forgot to add that to lbu as an exception 2015-06-26 14:41:28 <_ikke_> ah ok, i still have the link 2015-06-26 14:41:34 im experimenting with swclock, which is a part of openrc 2015-06-26 14:41:37 apparently does the same thing 2015-06-26 14:41:42 which means we wouldn't need a package. 2015-06-26 14:42:00 > Systems without a working real time clock (RTC) chip should use /etc/init.d/swclock , which sets the system time based on the mtime of a file which is created at system shutdown. 2015-06-26 14:42:05 but im not sure where that saves it's file 2015-06-26 14:42:22 an exception needs to be given to LBU if it's not in /etc 2015-06-26 14:46:19 i shall try this first. 2015-06-26 14:49:12 brb 2015-06-26 14:50:21 <_ikke_> sillysausage: http://alpine.ikke.info build dir (key is in there too) 2015-06-26 14:50:26 <_ikke_> sillysausage: fake-hwclock 2015-06-26 14:50:29 <_ikke_> haven't tested it 2015-06-26 15:07:46 i couldn't figure it out from here https://gitweb.gentoo.org/proj/openrc.git/tree/src/rc/swclock.c 2015-06-26 15:26:06 hmm _ikke_ i think we will need to use fake_hwclock 2015-06-26 15:26:26 the reason is because swclock touches 2015-06-26 15:26:28 lrwxrwxrwx 1 root root 12 Jan 1 1970 /sbin/openrc-run -> /sbin/openrc 2015-06-26 15:26:39 which ofcourse is going to always have the date of when it is packaged 2015-06-26 15:26:45 so some time in april 2015-06-26 15:26:56 where as fake-hwclock actually makes a file in /etc/ and puts the date in it 2015-06-26 15:47:48 <_ikke_> sillysausage: http://alpine.ikke.info build dir (key is in there too) 2015-06-26 15:48:24 okay :) 2015-06-26 15:48:32 <_ikke_> Haven't tested it yet 2015-06-26 15:48:42 yeah probably works i am currently experimenting with swclock 2015-06-26 15:48:45 but it's a pain in the ass 2015-06-26 15:50:30 <_ikke_> abuild automatically puts it in the x86_64 dir. Don't know if that causes problems for ARM 2015-06-26 16:31:59 hmm, why won 2015-06-26 16:32:02 't dnscrypt install now :P 2015-06-26 16:32:06 thinks libsodium is in @edge 2015-06-26 16:32:51 # apk add dnscrypt-proxy@testing 2015-06-26 16:32:54 ERROR: unsatisfiable constraints: 2015-06-26 16:32:56 libsodium-1.0.3-r0: 2015-06-26 16:32:58 masked in: @edge 2015-06-26 16:33:00 satisfies: dnscrypt-proxy-1.5.0-r0[so:libsodium.so.13] 2015-06-26 16:34:24 oh unless dnscrypt-proxy just got updated or something to require the libsodium from edge 2015-06-26 16:34:27 guess that's possible. 2015-06-26 16:42:46 anyone using alpine in a diskless setup? I'm interested in running a slimmed down version which I'd deploy over something like pxe or ipxe 2015-06-26 16:43:15 I had heard some rumors that there is a ~15-30mb version floating around somewhere 2015-06-26 17:01:52 _ikke_: so basically the same thing can be achieved with swclock, this is how i got it to work 2015-06-26 17:02:11 _ikke_: in case you want to know how to do it https://dpaste.de/PA9y 2015-06-26 17:02:35 _ikke_: you must also remember to "lbu include /etc/init.d/swclock" 2015-06-26 17:02:50 otherwise the one from the package will be unpacked and you wont be able to read your clock_file 2015-06-26 17:53:25 dlaube: I run an old eeeepc diskless as music device 2015-06-26 17:53:43 booting it via pxe 2015-06-26 17:54:28 jomat: nice! 2015-06-26 17:54:50 did you have to do anything special or are you running the full ~200mb os? 2015-06-26 18:02:32 why would the os be 200mb? 2015-06-26 18:35:39 dlaube: alpine isn't that big, is it? 2015-06-26 18:36:13 dlaube: but I can't look it up right now, I'm not at home and my flat seems to be offline since exactly 23.5 hours ._. 2015-06-26 18:36:27 usb image is 87 meg 2015-06-26 18:36:36 based off the mini iso 2015-06-26 19:31:10 ahh, I was looking at the regular installer iso 2015-06-26 19:31:24 no worries jomat —good luck with that! 2015-06-26 19:39:44 dlaube: and i think minimal bootstrapped installs without kernel ( -> containers ) are even >10MB afair 2015-06-27 03:13:25 hi 2015-06-27 05:05:22 hi guys, I just stumbled across this nice little distro because of docker, however I am searching for the paste utility, and can't seem to find it? is there a replacement? 2015-06-27 05:22:16 sorry, I was disconnected... 2015-06-27 05:22:27 ak5: paste utility? 2015-06-27 05:22:28 is there a replacement for the unix paste utility? 2015-06-27 05:22:53 oh 2015-06-27 05:22:58 like the old unix tool 2015-06-27 05:22:59 it's in coreutils 2015-06-27 05:23:32 can apk tell me this or any related tool? I am coming from archlinux 2015-06-27 05:24:00 ... I am honestly not sure 2015-06-27 05:24:05 http://pkgs.alpinelinux.org/contents?filename=paste&path=&pkgname=&arch=x86 this exists though 2015-06-27 05:24:20 presumably there's some way to 2015-06-27 05:24:22 ah cool 2015-06-27 05:24:34 contents of packages are searchable, I overlooked that - thanks! 2015-06-27 05:25:10 yeah it's really useful 2015-06-27 05:25:25 agreed 2015-06-27 05:32:47 and of course, refuge from systemd 2015-06-27 05:32:51 ACTION hides 2015-06-27 05:34:39 :&& 2015-06-27 05:35:17 well systemd makes a great operating system, lacking only a decent init system 2015-06-27 05:37:54 ...wait you stole that from vim vs emacs... 2015-06-27 05:41:08 parody falls under fair use! 2015-06-27 11:56:33 what's the right way to configure multiple of both ipv4 and ipv6 addresses on an interface in alpine? 2015-06-27 11:58:58 ah i see. can't have "iface eth0 inet6 static" as well as "iface eth0 inet static", the second address needs to be e.g eth0:0 2015-06-27 11:59:35 was getting 'RTNETLINK answers: File exists' in my console during boot 2015-06-27 12:00:50 Nemykal, http://wiki.alpinelinux.org/wiki/Configure_Networking#Example:_Dual-Stack_Configuration seems to say that you can and should 2015-06-27 12:01:17 yeah when I had it configured just like that, I had the aforementioned issue 2015-06-27 12:01:37 weird 2015-06-27 14:55:46 Nemykal - dual stack + dual ip's bridged http://hastebin.com/eqeheniqim.hs 2015-06-27 14:58:35 someone once posted a linkt to a hybrid install image 2015-06-27 14:58:43 please can someone repost that link? 2015-06-27 15:15:24 uploading a usb image now 2015-06-27 15:32:35 leo-unglaub - see http://it-offshore.co.uk/alpine/usb-mini - the image is compressed with xz so it can be accessed with libguestfs & edited / updated 2015-06-27 15:32:52 BitL0G1c: big thanks 2015-06-27 15:32:59 ;-) np 2015-06-27 15:33:20 i have no idea why setp-bootable never works for me 2015-06-27 15:37:23 that image s/be ok - I made a bootable usb with: 2015-06-27 15:37:51 xz -dc ./alpine-mini-3.2.0-x86_64-512M.img.xz | sudo dd bs=16M of=/dev/sda1 2015-06-27 15:38:32 & set the partition to be bootable with fdisk 2015-06-27 16:03:13 ahills, BitL0G1c, jzono1, clandmeter and anyone else who couldn't get apk to run inside chroot. 2015-06-27 16:03:26 Looks like the issue has been fixed: http://git.alpinelinux.org/cgit/apk-tools/commit/?id=68453b99e68e8bc8454956cdcbbf6d0a520d666f 2015-06-27 16:05:44 ^^^^ stateless - thank you 2015-06-27 16:20:00 BitL0G1c: https://github.com/itoffshore/alpine-linux-scripts/blob/master/setup-partitions#L29 maybe you could add here a check for all interfaces 2015-06-27 16:20:22 because on two of my maschines i have eth1 as the working inet device and then it always resetups the interfaces 2015-06-27 16:22:54 ok - will do - thx for picking up on this 2015-06-27 16:24:51 BitL0G1c: i love the new question for the luks params 2015-06-27 16:24:55 soooooo helpful!!!! 2015-06-27 16:25:51 that way i can use -s 512 -h sha512 insead of your defaults :) 2015-06-27 16:26:10 yes the encryption is now configurable via dialog - will do a video for it a bit later 2015-06-27 16:26:51 yeah, i love it :) 2015-06-27 16:27:16 I did quite a lot of reading on ciphers - your passphrase needs to have more than 128 bits of entropy on keys > 128 bits 2015-06-27 16:28:03 keepassX will show you how much entropy a passphrase has 2015-06-27 16:30:17 & in XTS mode -s 512 is actually a 256 bit key 2015-06-27 16:30:55 BitL0G1c: yeah, the entire ciphers topic is so fucking hard to get right ... 2015-06-27 16:31:18 thats one of those things that needs to be steamlines for crypto to really become mainstream 2015-06-27 16:31:38 because most it security guys with years of experience still get it wrong 2015-06-27 16:32:55 I added some code to prevent CBC mode being used with any IV except ESSIV 2015-06-27 20:13:57 anyone know who owns the dl-4 mirror, or even better, if there is a list of mirror owners somewhere? 2015-06-27 20:17:50 andyshinn: use whois 2015-06-27 20:19:59 whois on what? the IP? 2015-06-27 20:21:10 it is a CNAME for mail.alpinelinux.org so my best guess is that it is controlled by the Alpine team, but i'm not sure 2015-06-27 20:23:09 yeah, I'm not sure either 2015-06-27 21:40:34 fedora's static busybox seems to be less strange than debian's 2015-06-27 21:40:35 I'm not sure who thought it was a good idea to have dpkg built into their busybox 2015-06-27 21:42:06 :) 2015-06-27 21:42:43 it seems useful to have at least minimal versions of package utilities in busybox 2015-06-27 21:43:17 for example if you're using a bb-based recovery disk to fix a hosed distro, you might need to manually extract packages for the distro over top of the fs 2015-06-27 21:43:38 not sure why you can't just include a staticly-linked dpkg to go with it 2015-06-27 21:43:56 surely the effort to 2015-06-27 21:44:59 assuming it's not a debian-specific rescue disk 2015-06-27 21:45:16 and i seriously doubt it's easy to make the debian program work without a whole debian ecosystem 2015-06-27 21:45:21 and it probably needs perl installed :) 2015-06-27 21:45:26 dpkg built into busybox though 2015-06-27 21:47:17 i guess i'm not understanding what you're saying 2015-06-27 21:47:59 https://fiche.greata.uk/ihjh https://fiche.greata.uk/0n7t https://fiche.greata.uk/9p4x 2015-06-27 21:48:36 those are respectively, some munged output from debian busybox-static, fedora busybox (also statically linked), and a diff 2015-06-27 21:53:58 wait fedora's has the whole runit suite baked in 2015-06-27 21:54:01 I am so confused 2015-06-27 21:57:29 wait that's actually a build option in upstream busybox isn't it 2015-06-27 22:13:47 Does the Tor Browser Bundle work on Alpine? 2015-06-27 22:14:01 According to Tor Developers, using tor with any other browser is insecure. 2015-06-27 22:14:02 https://www.torproject.org/docs/faq.html.en#TBBOtherBrowser 2015-06-27 22:14:31 it's also insecure with the tbb... 2015-06-27 22:14:55 how so? 2015-06-27 22:15:26 how not so? 2015-06-27 22:15:59 firefox as such is a pile of bugs 2015-06-27 22:16:23 security by design is something very different. 2015-06-27 22:16:38 the tor people can patch it as much as they want, it remains a pile of crap 2015-06-27 22:16:56 at least it's not sabotaged so much as chromium though from a privacy perspective 2015-06-27 22:17:06 still safer than using default firefox and tor (only available method right now on alpine, I think) 2015-06-27 22:17:20 tbb is better, even though it isn't perfect (far from it) 2015-06-27 22:17:22 well, why don't you use tails then? 2015-06-27 22:17:24 still the best we have 2015-06-27 22:17:46 bCQc0P3DIqbh: too lazy to reboot all the time 2015-06-27 22:18:06 well. too lazy to do opsec. so why do you care then? 2015-06-27 22:18:07 not that I need tor, but it's good to have it as an option on alpine, just in case 2015-06-27 22:18:32 opsec is damn hard. rebooting is just one aspect 2015-06-27 22:19:44 unless you're doing something illegal, there's no need to worry that much. If I simply want to visit a website without letting the owner know my IP or something like that, there's no need to worry about all details of opsec and simply running tor will be enough 2015-06-27 22:19:56 don't you agree? it's not like security is binary 2015-06-27 22:20:05 there are different levels of security for different needs 2015-06-27 22:20:14 threat modelling it's called 2015-06-27 22:20:23 but most threats include the state now ;/ 2015-06-27 22:21:30 still, tbb offers much more security than firefox+tor and it would be cool to have the tor browser bundle working on alpine 2015-06-27 22:21:32 that's all... 2015-06-27 22:21:37 because most states are criminals themselves. the biggest ones. 2015-06-27 22:22:03 if you only visit a random sme/private persons website, then tor with firefox is probably enough 2015-06-27 22:22:23 if you visit one of the drugmarket honeypots, then not even tbb will save your ass 2015-06-27 22:23:00 if your adversary is an aol user, you can even use iexplorer i guess ;) 2015-06-27 22:23:01 yeah, I guess so... depends on the use case 2015-06-27 22:23:10 yeah, lol 2015-06-27 22:50:14 leo-unglaub - setup-partitions s/be fixed now for multiple interfaces https://github.com/itoffshore/alpine-linux-scripts/commit/2916c9cf3c51fd2a92257598b55a3e1f75e4e128 - it only touches interfaces if running route gives an error 2015-06-27 22:51:33 BitL0G1c: ah, thats a nice idea 2015-06-27 22:51:52 i would have done a for over /sys/class/net/ and checked every one 2015-06-27 22:51:56 but yours is much shorter :) 2015-06-27 22:52:01 thanks for implementiung it 2015-06-27 22:52:16 np - thanks for testing 2015-06-27 22:52:41 yes i thought of that too first - but seemed simpler just to check route for an error 2015-06-27 22:53:52 is there a special rule in apline linux that disallows suid on an executable? 2015-06-27 22:54:38 http://dpaste.com/2MYZPDT.txt 2015-06-27 22:54:49 why is that an error? 2015-06-27 23:05:14 wait does apk fetch check sigs or should I verify it separately 2015-06-27 23:06:44 sigs are verified against keys in /etc/apk/keys - if you ever build an apk manually you need to add your public key there or you can only install an apk manually with --force 2015-06-27 23:07:57 apk add will error not apk fetch 2015-06-27 23:09:50 actually looks like fetch will error too --allow-untrusted 2015-06-28 03:46:01 hm how do I tell if a given subdir of /proc is a pid or the other stuff 2015-06-28 04:08:56 ammunta: If the directory name consists entirely of digits, then it's a pid, otherwise it isn't. You can do that test with the proper regex and echo $filename |grep -q 2015-06-28 08:21:36 how to find what process is running port 9000? 2015-06-28 08:21:50 i am trying to run an angular app on that port it doesn't let me 2015-06-28 08:22:04 ps gives me just 2 lines.... 2015-06-28 08:22:37 lsof -i:9000 gave me nothing 2015-06-28 08:25:47 debug your app 2015-06-28 08:28:56 netstat -lptun 2015-06-28 08:29:09 as root 2015-06-28 09:36:54 BitL0G1c: thanks 2015-06-28 12:49:30 has someone of you started creating packages for the gnunet? 2015-06-28 12:49:34 if not, i am getting on it 2015-06-28 19:30:24 issues with installing packages: aterm (missing): required by world[aterm] 2015-06-28 19:30:51 <_ikke_> means apk does not know about a package called aterm 2015-06-28 19:31:35 but 2015-06-28 19:31:48 wiki says i should use it: http://wiki.alpinelinux.org/wiki/Awesome%28wm%29_Setup 2015-06-28 19:31:56 or should i ignore wiki? 2015-06-28 19:32:45 <_ikke_> I think aterm is just an example terminal 2015-06-28 19:33:32 <_ikke_> I cannot find it 2015-06-28 19:33:46 <_ikke_> It was last in 2.6 2015-06-28 19:34:06 <_ikke_> So the wiki is probably outdated 2015-06-28 19:34:56 hm, thanks 2015-06-28 19:35:02 xterm is fine too 2015-06-28 19:35:23 RLa: if you don't need aterm explicitly, just use something like rxvt-unicode instead 2015-06-28 19:36:25 i'm having issues getting X running at all 2015-06-28 19:36:36 oh 2015-06-28 19:36:51 startx is not found, so awesome does not install xorg as dependency? 2015-06-28 19:37:09 apk works like slackware packages, no dependency tracking? 2015-06-28 19:37:13 awesome just could use a remote x-server 2015-06-28 19:37:14 a WM doesn't depend on an X server 2015-06-28 19:37:20 what jomat said 2015-06-28 19:37:26 hm, indeed 2015-06-28 19:37:57 <_ikke_> You don't want unnecessary dependencies 2015-06-28 19:38:02 /usr/bin/startx is owned by xinit-1.3.3-r3 2015-06-28 19:38:47 also a common problem: did you run setup-xorg-base or manually replace dbus? 2015-06-28 19:38:59 i'm a long time slackware user (since version 8) and used to have most things installed :) 2015-06-28 19:40:18 but yeah, its install has over 4GB apps, most that i never use 2015-06-28 19:43:46 how popular is running xfce4 on alpine? 2015-06-28 19:44:13 maybe i'm only one trying to install/run awesome? 2015-06-28 19:44:59 most feedback i heard was re i3wm here 2015-06-28 19:45:41 i think i'm too much of a mouse user to use a tiling wm :) 2015-06-28 19:46:02 i use xfce4 2015-06-28 19:47:20 alpine is rolling-release or fixed-release? 2015-06-28 19:47:49 edge is available and works well for me, but there are fixed releases too 2015-06-28 22:09:37 ahoi 2015-06-29 00:34:56 kocka: ohai o/ 2015-06-29 01:56:36 RLa: setup-xorg-base 2015-06-29 01:56:59 And some devs use xfce4 I think 2015-06-29 01:57:15 Oh, oops 2015-06-29 02:46:39 ahills: btw: your clock is very wrong 2015-06-29 02:47:50 Your mail regarding hibernation was sent on Mon, 29 Jun 2015 14:59:37 +0000, but I received it on 27 Jun 2015 19:00:46 -0000 2015-06-29 02:49:15 Just noticed it because it keeps staying the first one in my inbox %) 2015-06-29 02:51:05 .oO(or did I forget to work the last two days... wouldn't bother neither) 2015-06-29 07:06:09 morning campers 2015-06-29 12:58:10 I have really weird problem with rsync on alpine erroring out on: failed to set permissions on "…": No such file or directory (2) on files that clearly exist 2015-06-29 12:59:55 I've just tried to deploy few alpine vservers (my scripts for that use rsync heavily) and it rather non-deterministically throws errors 2015-06-29 13:01:01 this is on edge 2015-06-29 13:03:49 CcxCZ: could it be that file is moved/reneamed while rsync is running? 2015-06-29 13:04:27 if you have a stable way to reproduce, then you can try run it strace 2015-06-29 13:07:14 Nope, nothing should be touching those files yet. It errors on about half of the files, but it seems random/inconsistent. They all get transferred properly though, only permission setting fails. 2015-06-29 13:08:02 I'll try to find less complex way to reproduce it. 2015-06-29 13:11:41 what filesystem is it? 2015-06-29 13:14:17 xfs, using non-chrooted (gentoo/glibc) rsync works flawlessly on the same location 2015-06-29 13:15:13 I'm using OpenSSH's ChrootDirectory to run things inside the alpine vserver prior it's startup 2015-06-29 13:16:08 can you reproduce it with strace? 2015-06-29 13:16:16 strace rsync .... 2015-06-29 13:23:33 Hmm would do some trickery to run the strace on remote-side, doable with some script. Anyway, now it ran correctly, I'll report back if I manage to get a sensible trace. 2015-06-29 13:26:06 I'm not sure if this only happens due to differing systems (local gentoo, remote alpine), but I think it errored on me just doing local operations on alpine too, tried to reproduce that and it's okay and nothing more to sync. 2015-06-29 13:29:26 it sounds like some sort of race 2015-06-29 13:30:08 could it be multiple rsync runs at same time? 2015-06-29 13:30:26 the first time it looked that way (errored on the tmpfiles) but then it just tries to alter existing files' perms … and fails randomly 2015-06-29 13:30:58 weird 2015-06-29 13:51:19 Is there an env.d / env-update equivalent? Or what is the preferred way to extend PATH? 2015-06-29 13:56:39 . /etc/profile ? 2015-06-29 13:56:59 i suppsoe just edit /etc/profile would do 2015-06-29 13:58:23 The nice thing about env.d is that it handles concatenating the PATH / ROOTPATH variables, but doing I can do it by hand like that too. 2015-06-29 15:13:30 what package provides dns tools for alpine line dig? 2015-06-29 15:13:33 like* 2015-06-29 15:14:56 ekristen: drill 2015-06-29 15:16:47 thanks 2015-06-29 15:16:52 apk info -W /usr/bin/dig [1] 2015-06-29 15:16:53 /usr/bin/dig is owned by bind-tools-9.10.2_p1-r1 2015-06-29 15:17:23 oh nice, I was looking for a command like that ScrumpyJack thanks 2015-06-29 15:17:25 you can also check pkgs.alpinelinux.org 2015-06-29 15:19:49 ScrumpyJack: that only works when the pkg is actually installed. 2015-06-29 15:24:57 clandmeter: indeed, and as i have dig, i just copy/pasted the result of --who-own :) 2015-06-29 19:52:59 Hi! I downloaded the iso and dd'ed it to a usb memory but it didn't boot. Do I have to use a CD? 2015-06-29 19:53:26 The wiki text is confusing/funny on this. 2015-06-29 19:55:15 First: "If you have a CD drive from which you can boot, then burn the ISO onto a blank CD using your favorite CD burning software. Else create a bootable USB drive." 2015-06-29 19:56:23 But to create a bootable USB, an Alpine Linux CD-ROM is required... 2015-06-29 20:00:24 Shouldn't it just work with CD or USB? 2015-06-29 20:07:34 Seems it's not a hybrid iso. 2015-06-29 20:16:36 <_ikke_> niko__: Correct 2015-06-29 20:16:45 <_ikke_> use something like unetbootin to make a bootable usb 2015-06-29 20:23:27 _ikke_, thanks for the tip. I'll do that if manually installing isolinux and copying the files to usb doesn't work. I'm almost done... 2015-06-29 20:36:14 niko__ install in vmware player or something like that an alpinelinux installation and do http://wiki.alpinelinux.org/wiki/Create_a_Bootable_USB#Manual 2015-06-29 21:23:11 niko__ see http://it-offshore.co.uk/linux/alpine-linux/48-alpine-linux-usb-hybrid-image 2015-06-30 00:31:43 bontibon: drill does in principle the same thing but is remarkably useless when what it turns out you want is +short 2015-06-30 08:08:48 so, I have getty running on some ttys (default inittab in AL) and I can switch ttys until I startx. once X is started, i lose the tty switching (well, I see the black tty, but no getty/login). when i kill X, still only black ttys. what gives? :) 2015-06-30 08:48:52 w 2015-06-30 13:22:09 ncopa: Okay, so I got the strace for the rsync errors. http://wpr.cz/ccx/paste/2015-06-30/0/raw.txt http://wpr.cz/ccx/paste/2015-06-30/1/ apparently it fails on all files, it was just interleaving of messages that confused me. I suspect it might be /proc being missing causing it. 2015-06-30 13:26:15 newfstatat(AT_FDCWD, "/proc/self/fd/0", 0x7fff0645a450, 0) = -1 ENOENT (No such file or directory) 2015-06-30 13:26:19 is it in a chroot? 2015-06-30 13:26:50 its /proc not beeing mounted 2015-06-30 13:27:16 Yeah, seems like the /proc is the issue. Now the only mystery is why only alpine is affected. 2015-06-30 13:28:05 there are libc calls that depend on /proc 2015-06-30 13:30:06 quite possibly can be the difference between glibc and musl 2015-06-30 13:32:39 I wonder if it's a reasonable thing to mount /vservers/foo/proc from the main fstab then. 2015-06-30 13:34:04 doest vserver provide some sort of /proc? 2015-06-30 13:36:19 Yes, the guests have separate fstab files for their separate namespaces. But I don't want to go into vserver context to set it up, among other reasons because I can't when it's not running. 2015-06-30 13:36:41 vprocunhide 2015-06-30 13:37:20 iirc you need run vprocunhie on the vserver host 2015-06-30 13:37:24 or similar 2015-06-30 13:38:01 i think you can config what to hide/unhide in your util-vserver config 2015-06-30 13:45:49 That I have. Thing is this is plain chroot since I use this for setting it up before I can start the actual vserver. This approach worked well with other systems, but oh well, I guess I can mount extra proc somehow. 2015-06-30 13:54:03 ps, top and other things will not work without /proc too 2015-06-30 13:55:40 ScrumpyJack: do you have the fbcon kernel module loaded? i think you need that to switch from X11 to tty 2015-06-30 13:56:18 Indeed. At this stage I only care about rsync & chown though. 2015-06-30 13:56:52 pnutzh4x0r: well spotted. makes sense. I'll check 2015-06-30 14:13:12 pnutzh4x0r: yup, that was the problem 2015-06-30 14:21:08 CcxCZ for an another problem i shoudl check if i can strace errors, do you have any information how to use strace? i know google is my friend but maybe you have some small doku for how to use it 2015-06-30 14:21:18 as it seems gdb cannot get everything 2015-06-30 14:23:25 well, gdb is more powerful than strace, but in 99% of cases strace is more convenient, for sysadmin at least. 2015-06-30 14:24:07 The common usage is just strace -f yourapplication; other flags of note are -e and -p. 2015-06-30 14:35:18 anyone running alpine-xen as their dom0? 2015-06-30 14:36:35 <- normal sys install 2015-06-30 15:35:12 ncopa: yup 2015-06-30 15:35:19 ncopa: many 2015-06-30 15:35:47 wassup? 2015-06-30 15:56:41 / # apk update 2015-06-30 15:56:41 fetch http://dl-4.alpinelinux.org/alpine/v3.2/main/x86_64/APKINDEX.tar.gz 2015-06-30 15:56:41 ERROR: http://dl-4.alpinelinux.org/alpine/v3.2/main: temporary error (try again later) 2015-06-30 15:56:43 WARNING: Ignoring APKINDEX.abd86498.tar.gz: No such file or directory 2015-06-30 15:56:45 1 errors; 15 distinct packages available 2015-06-30 15:56:47 Hello there, I have recently started using alpine linux as my base image in all my docker containers and I sometimes run into issues with apk. The following is an error I just got when running apk update: 2015-06-30 15:57:47 https://gist.github.com/anonymous/85c22f4d0c713e33f549 2015-06-30 15:57:57 that might be easier to read 2015-06-30 15:58:34 Am I doing something wrong or is there another mirror I can be using for apk or something? 2015-06-30 15:58:42 <_ikke_> cwandrews: can you ping dl-4.alpinelinux.org? 2015-06-30 16:00:05 bad address, so I assume that my boot2docker vm's dns is screwed up which happens kind of often 2015-06-30 16:00:10 sorry for the noise 2015-06-30 16:21:56 yeah it was dns issue in boot2docker everything is working fine. I love alpine linux. keep up the good work! 2015-06-30 16:22:53 <^7heo> I am using edge 2015-06-30 16:23:05 <^7heo> I try to connect to a bluetooth device 2015-06-30 16:23:25 <^7heo> and I get errors on "connect " 2015-06-30 16:23:34 <^7heo> strace shows: recvmsg(4, 0x7cca1c7d7cd8, MSG_CMSG_CLOEXEC) = -1 EAGAIN (Resource temporarily unavailable) 2015-06-30 16:24:58 <^7heo> I'm posting that here because maybe someone already has encountered that issue, and has a solution. 2015-06-30 16:34:02 <^7heo> okay it seems to rely (at least partly) on dbus 2015-06-30 16:34:07 <^7heo> and has files from systemd. 2015-06-30 16:34:19 <^7heo> I can imagine why it isn't working. 2015-06-30 18:12:56 clandmeter: what was the sensors issue you had ? 2015-06-30 18:13:44 "sensors-detect"* 2015-06-30 18:23:31 I am unable to get "ACPI interface" info on Gigabyte-H81M-DS2 2015-06-30 18:57:24 hmmm... my second AL bare metal up and running !! 2015-06-30 18:57:41 but yet to decide what to use it for 2015-06-30 21:15:01 In a freshly built alpine base image, I installed update-ca-certificates and tried to run it, but got this error message: "WARNING: ca-certificates.crt does not contain exactly one certificate or CRL: skipping" 2015-06-30 21:15:35 I saw this when attempting to add my self-signed cert to the container, so I tried running update-ca-certificates without modifying anything on the system and also saw that error 2015-06-30 21:16:58 sbuss: which alpine release? 2015-06-30 21:17:09 latest, and 3.1 2015-06-30 21:17:13 haven't checked others 2015-06-30 21:17:28 i thought this was fixed in 3.2 2015-06-30 21:17:36 atlease i think ive seen this error before 2015-06-30 21:18:07 I found a couple hits on google, but no solution 2015-06-30 21:19:55 reproduce with 2015-06-30 21:19:55 FROM alpine:latest 2015-06-30 21:19:55 RUN apk --update add ca-certificates 2015-06-30 21:19:55 RUN update-ca-certificates 2015-06-30 21:19:59 maybe the fix is in edge? http://git.alpinelinux.org/cgit/aports/log/main/ca-certificates 2015-06-30 21:20:13 is that tagged as `:edge`? 2015-06-30 21:20:23 master is edge 2015-06-30 21:21:50 I mean is there a docker image tagged as edge? looks like it, but I'm not sure how closely that tracks master 2015-06-30 21:23:10 ah, it occurs to me that https://github.com/gliderlabs/docker-alpine may not be part of the alpine linux project and just exists as a convenience 2015-06-30 21:27:14 Yea, edge is master 2015-06-30 21:35:23 I just built the docker-alpine images locally and it still fails with the same error 2015-06-30 21:36:41 sbuss: what is the actual error? That is just s warning 2015-06-30 21:40:48 I suppose I might be doing the cert stuff incorrectly, but this exact same set of commands works on ubuntu 2015-06-30 21:42:49 https://www.irccloud.com/pastebin/zhjo7o5c/alpine%20update-ca-certificates 2015-06-30 21:42:56 that's my whole Dockerfile ^^ 2015-06-30 21:43:27 then `docker run -t -i c72c2b1a2261 curl -v https://authmanager.counsyl.com` shows that curl doesn't trust the cert 2015-06-30 21:43:41 so that Warning indicates it's not actually updating the system ca-certs 2015-06-30 21:44:21 (authmanager.counsyl.com is on our internal network, using our internal self-signed cert) 2015-06-30 21:57:00 sbuss: you have to drop it into /etc/ssl/certs too 2015-06-30 21:59:41 sbuss: there was also a bug in the update-ca-certificates which i think was only fixed in 3.2, so be sure you are FROM alpine:3.2 2015-06-30 22:01:16 putting it in /etc/ssl/certs had no effect 2015-06-30 22:01:34 andyshinn: I'm using a locally built alpine container based off of master@HEAD 2015-06-30 22:01:44 but it also fails with alpine:3.2 2015-06-30 22:01:55 even if I don't add my own certs, that command fails 2015-06-30 22:02:07 can you verify the symlink to the cert does get added to /etc/ssl/certs? 2015-06-30 22:02:51 it should be find in /usr/local/share/ca-certificates 2015-06-30 22:03:19 i had this same issue before and it was a bug in the update script, so i know the script should work in this exact scenario 2015-06-30 22:04:02 i wonder if the update script has trouble handling subdirs though, have you tried adding the cert directly to /usr/local/share/ca-certificates/? 2015-06-30 22:04:27 after running the script, you should get a symlink to it in /etc/ssl/certs 2015-06-30 22:04:34 I'll try that now 2015-06-30 22:07:20 seems unlikely it'd have an issue with subdirs, since /etc/ca-certificates.conf references a bunch of certs in directories 2015-06-30 22:08:41 but the code specifically globs on *.crt in that folder: https://github.com/alpinelinux/aports/blob/3.2-stable/main/ca-certificates/update-ca-certificates#L47 2015-06-30 22:08:50 which wouldn't be subdirs 2015-06-30 22:09:19 i wonder if lua supports something like **/*.crt 2015-06-30 22:09:46 hm 2015-06-30 22:09:49 interesting 2015-06-30 22:09:57 I think /etc/ca-certificates.conf has no effect? 2015-06-30 22:10:59 that is probably for stuff that works with the openssl utilities... i'm not really sure 2015-06-30 22:11:33 i think most apps just grab the /etc/ssl/certs/ca-certificates.crt file 2015-06-30 22:12:00 hm, that's not the right filename: ca-cert-counsyl-root-ca.pem.pem 2015-06-30 22:13:17 i'm just going based on your example, i don't see where you reference a pem file 2015-06-30 22:13:36 yeah, sorry, that was a local mistake 2015-06-30 22:13:42 not related to the core issue 2015-06-30 22:15:05 but it i worth noting the code to see how it works, one issue i ran into was naming the files .pem and .cert at one point, which also will not work since it specifically globbed on .crt 2015-06-30 22:15:57 ok it turns out that `/usr/local/share/ca-certificates` is, in fact, not an acceptable location for certs 2015-06-30 22:17:40 woohoo "* SSL connection using TLSv1.2 / ECDHE-RSA-AES256-GCM-SHA384" 2015-06-30 22:18:11 https://www.irccloud.com/pastebin/fxkb8K5c/working%20alpine%20update-ca-certificates%20with%20self-signed%20cert 2015-06-30 22:18:19 that's my working Dockerfile ^ 2015-06-30 22:18:27 so alpine:3.2 works fine, turns out 2015-06-30 22:19:10 hey, works with 3.1 as well 2015-06-30 22:19:15 so it was just a path issue :facepalm: 2015-06-30 22:19:18 thanks for the help, all 2015-06-30 22:20:23 oh haha, that would be an issue