2022-04-01 03:39:38 Man, I spent about 10 minutes trying to find options. There are two things that *might* help - the 'netdev' and 'ether' (deprecated by netdev) cmdline params. The one solution I know would work right now is to blacklist the driver on the kernel cmdline: .blacklist=yes 2022-04-01 03:40:13 Pardon me - it's actualy: modprobe.blacklist= 2022-04-01 03:40:38 @smokeyhamlin 2022-04-01 08:41:36 ACTION uploaded an image: (121KiB) < https://matrix.org/_matrix/media/r0/download/matrix.org/WdIplNCHIaJELUzXWShFOnTW/image.png > 2022-04-01 08:41:37 morning, party people 2022-04-01 08:41:37 i'm installing Alpine manually and trynna mount the partitions (formatted ext4), but the mount tool fails 2022-04-01 08:42:01 how can i troubleshoot it further or solve it? 2022-04-01 08:44:35 the ISO is 3.15.3, i've been following all the instructions from the docs 2022-04-01 08:49:10 mount -t ext4 /dev/sda3 /mnt should work 2022-04-01 08:57:30 it worked, cheers 2022-04-01 08:59:55 and for the ESP it's -t "vfat"? 2022-04-01 09:00:02 * and for the ESP it's "-t vfat"? 2022-04-01 09:05:28 that or fat32 2022-04-01 09:55:05 i've installed the system and the swap isn't turned on automatically. did i miss the flag in the "setup-disks /mnt"? (i know how to add swap manually) 2022-04-01 09:55:31 acid-bong[m]: Were there any errors while running setup-disk? 2022-04-01 09:55:53 "fix MBR" 2022-04-01 09:56:11 which wasn't present yesterday, when i ran setup-alpine 2022-04-01 09:57:24 the difference i noticed is that VirtualBox recognized only the automatically installed system, not the manual one 2022-04-01 09:57:39 Hmm, because I thought I remembered there was some problem with it not properly formatting partitions, but I'm not quite sure how that could have been happening since it's specifying a mountpoint. 2022-04-01 09:57:48 but when i choose the correct drive in the UEFI menu, it boot normally 2022-04-01 09:57:52 s/boot/boots/ 2022-04-01 09:57:52 acid-bong[m] meant to say: but when i choose the correct drive in the UEFI menu, it boots normally 2022-04-01 17:12:01 hi? 2022-04-01 17:17:39 Hello 2022-04-01 17:18:27 how are you? 2022-04-01 17:28:48 bye bye 2022-04-01 17:28:48 __ _| |_ __ (_)_ __ ___ (_)___ | |_| |__ ___ | |__ ___ ___| |_ 2022-04-01 17:28:48 _ _ _ _ _ _ _ 2022-04-01 17:28:50 | (_| | | |_) | | | | | __/ | \__ \ | |_| | | | __/ | |_) | __/\__ \ |_ 2022-04-01 17:28:50 / _` | | '_ \| | '_ \ / _ \ | / __| | __| '_ \ / _ \ | '_ \ / _ \/ __| __| 2022-04-01 17:28:52 \__,_|_| .__/|_|_| |_|\___| |_|___/ \__|_| |_|\___| |_.__/ \___||___/\__| 2022-04-01 17:28:52 |_| _ _ _ _ _ _ _ 2022-04-01 17:28:54 / _` | | '_ \| | '_ \ / _ \ | / __| | __| '_ \ / _ \ | '_ \ / _ \/ __| __| 2022-04-01 17:28:54 __ _| |_ __ (_)_ __ ___ (_)___ | |_| |__ ___ | |__ ___ ___| |_ 2022-04-01 17:28:56 | (_| | | |_) | | | | | __/ | \__ \ | |_| | | | __/ | |_) | __/\__ \ |_ 2022-04-01 17:28:56 \__,_|_| .__/|_|_| |_|\___| |_|___/ \__|_| |_|\___| |_.__/ \___||___/\__| 2022-04-01 17:28:58 |_| 2022-04-01 17:29:00 _ _ _ _ _ _ _ 2022-04-01 17:29:02 __ _| |_ __ (_)_ __ ___ (_)___ | |_| |__ ___ | |__ ___ ___| |_ 2022-04-01 17:29:04 / _` | | '_ \| | '_ \ / _ \ | / __| | __| '_ \ / _ \ | '_ \ / _ \/ __| __| 2022-04-01 17:29:05 can you not 2022-04-01 17:29:07 | (_| | | |_) | | | | | __/ | \__ \ | |_| | | | __/ | |_) | __/\__ \ |_ 2022-04-01 17:29:27 lol 2022-04-01 20:50:30 Hello, On my setup , foot and alacritty have very dull colors 2022-04-01 20:58:38 have you tried changing them 2022-04-01 21:55:01 is there a way to easily duplicate grub menu entry, need different UUID but everything else same and in sync 2022-04-02 01:14:49 :wqa 2022-04-02 01:15:02 Well that didn't work. 2022-04-02 20:18:07 hello, in the wiki i've found this line: 2022-04-02 20:18:13 armhf: The newer ARM hard-float for newer, more powerful 32-bit devices alongside 64-bit 2022-04-02 20:19:31 does it mean that there are either 32bit or 64bit devices, or does it mean that 32bit can run alongside 64bit, which would mean that there is multilib support? 2022-04-02 20:19:45 with the latter i mean running 32bit binaries on 64bit arm on alpine linux for instance 2022-04-02 20:25:10 armhf is only 32-bits 2022-04-02 20:25:44 there are aarch64 cpus that can run in 32-bits mode 2022-04-02 20:27:18 ok, but is it possible to have both 32 bit and 64 bit gcc on a alpine-linux arm installation? 2022-04-02 20:27:32 no problem with /lib interference? 2022-04-02 20:27:44 alpine has no multi-lib support atm 2022-04-02 20:28:03 what's the usual workaround? a chroot? 2022-04-02 20:29:15 for example 2022-04-02 20:30:15 why am I not surprised by the fact who wrote that 2022-04-02 20:30:46 Picoro? 2022-04-02 20:31:16 a.k.a mckaygerhard 2022-04-02 20:31:21 nod 2022-04-02 20:53:49 Ah, this is the guy who administers alpine telegram chats 2022-04-02 20:54:30 "administers" might not exactly reflect the state of that place 2022-04-02 21:19:59 Hehe 2022-04-02 21:49:06 "The newer ARM hard-float for newer, more powerful 32-bit devices alongside 64-bit. Including video games!" 2022-04-02 21:49:24 they put video games in my armhf 2022-04-02 21:53:18 wiki page updated, now with more English and less video games 2022-04-02 22:05:03 Wee! 2022-04-03 01:10:21 anyone have foreign input method using ibus or fcitx5 on alpinelinux+xfce4 successfully? 2022-04-03 03:16:16 No, but I probably should try it... What language are you trying? 2022-04-03 10:24:48 do we need to setup something on GNOME to remember ssh keys? it works on every distributions but on alpine for some reasons it keeps asking me to type the key 2022-04-03 10:25:58 i try to keep gui specific stuff like that out of it 2022-04-03 10:26:21 are you using an agent already? 2022-04-03 10:28:53 seems there are no ssh agent environment variable, I wonder if systemd does something magic on other distributions or if they some autostart stuff 2022-04-03 10:29:40 no SSH_AUTH_SOCK? 2022-04-03 10:29:55 ah yes there is one 2022-04-03 10:30:54 what is it? 2022-04-03 10:31:11 SSH_AUTH_SOCK=/tmp/xdg/1000/keyring/ssh 2022-04-03 10:31:31 xdg, so it could be some desktop thing that set it up 2022-04-03 11:03:31 its gnome-keyring 2022-04-03 11:05:37 export SSH_AUTH_SOCK=(gnome-keyring-daemon --components=pkcs11,secrets,ssh --start --daemonize 2> ~/.local/var/log/gnome-keyring-daemon.log | string split '=') 2022-04-03 11:06:18 (this is for fish, you need to tweak it for bash/zsh/ash) 2022-04-03 11:06:30 (and drop whichever options you don't want) 2022-04-03 11:06:56 and put it in your shell profile 2022-04-03 11:07:37 panekj, thanks I'll look at it 2022-04-03 11:10:01 if you need/want git to take stuff out of keyring or just generally use secret service API, there is package (in edge) git-credential-libsecret 2022-04-03 11:11:04 other secret service provider is keepassxc (needs to be enabled in options, cannot coexist with gnome-keyring) 2022-04-03 11:44:30 Are there any default limitations to kernel message buffer size? Dmesg doesn't retrieve messages which appeared at first 0.3 seconds 2022-04-03 12:04:15 its ciruclar aswell 2022-04-03 12:04:50 also make sure to be root when looking at dmesg 2022-04-03 12:04:56 ubuntu changed that some time ago 2022-04-03 12:05:05 but I see from 0.000 when microcode is being loaded 2022-04-03 12:08:24 I can see everything in ubuntu, but not in alpine 2022-04-03 12:09:46 yeah I can confirm that 2022-04-03 12:09:58 first entry is from about 0.340792 in alpine 2022-04-03 12:13:27 I think it should be some kind of kernel param or something 2022-04-03 12:14:36 kernel compile MSGBUFSIZE seems to be a limit during compile 2022-04-03 12:17:58 log_buf_len=n[KMG] Sets the size of the printk ring buffer, 2022-04-03 12:20:59 where did you find this? 2022-04-03 12:21:08 internet 2022-04-03 12:21:20 look at /boot/config-virt 2022-04-03 12:21:22 search for LOG 2022-04-03 12:21:37 alpine seems to default to 2022-04-03 12:21:47 LOG_BUG_SHIFT=14 2022-04-03 12:21:59 LOG_CPU_MAX_BUF_SHIFT=12 2022-04-03 12:22:15 PRINTK_SAFE_LOG_BUF_SHIFT=13 2022-04-03 12:22:35 https://unix.stackexchange.com/questions/541945/sysctl-a-doesnt-include-log-buf-len 2022-04-03 12:24:03 ill try 19 and see how it behaves 2022-04-03 12:28:19 Apachez: yay, it worked! 2022-04-03 12:31:28 it seems to be to add that as default param in /etc/update-extlinux.conf 2022-04-03 12:31:37 and then as root run /sbin/update-extlinux and reboot 2022-04-03 12:32:07 but 2022-04-03 12:32:08 hm 2022-04-03 12:32:11 what did you set it to? 2022-04-03 12:32:21 I tried 19 but now I still dont see from 0.000 2022-04-03 12:33:03 I set it to 524288 (i. e. 2^19) 2022-04-03 12:33:28 LOG_BUF_SHIFT=524288 ? 2022-04-03 12:33:36 or LOG_BUF_LEN ? 2022-04-03 12:33:53 log_buf_len 2022-04-03 12:34:27 log_buf_shift takes binary logarithm of desired buffer length 2022-04-03 12:38:09 hmm it didnt like alterning shift 2022-04-03 12:38:12 ill try _len 2022-04-03 12:38:31 len it is :) 2022-04-03 12:38:41 as root: 2022-04-03 12:38:49 vi /etc/update-extlinux.conf 2022-04-03 12:39:09 append to default_kernel_opts: log_buf_len=1M 2022-04-03 12:39:10 save 2022-04-03 12:39:24 run: /sbin/update-extlinux 2022-04-03 12:39:40 verify with cat /boot/extlinux.conf 2022-04-03 12:39:41 reboot 2022-04-03 12:39:48 then dmesg | less and tada! :) 2022-04-03 12:40:23 Yeah, I did it 2022-04-03 12:41:00 seems like ubuntu defaults to 256k so that should be enough 2022-04-03 12:47:08 Bigger buffer lenght = heavier kernel 2022-04-03 12:47:42 s/lenght/length/ 2022-04-03 12:47:42 Ermine meant to say: Bigger buffer length = heavier kernel 2022-04-03 12:48:00 but 256kbyte ram isnt that heavy 2022-04-03 12:48:08 but sure 2022-04-03 13:31:09 it is on some machines 2022-04-03 13:32:16 if you only have 8 MB then 0.25 MB for a single log is a lot 2022-04-03 16:26:37 if you only have 8MB then you would disable dmesg all together :P 2022-04-03 16:34:54 that would make debugging impossible on most openwrt devices 2022-04-03 16:38:16 only seeing 200 rows is also virtually impossible to debug 2022-04-03 16:38:44 and if you are in such limited environment then you should change that setting 2022-04-03 16:38:49 not the other way around 2022-04-03 17:34:22 Hello71: openwrt doesn't support 8MB devices anymore afaik 2022-04-03 17:34:33 sort of 2022-04-03 17:41:10 But I think increasing LOG_BUF_SHIFT to 15 shouldn't hurt much (this corresponds to 32 KB buffer) 2022-04-03 17:47:22 but I can agree that by default it shouldnt be too large 2022-04-03 17:47:32 but right now its too small when you cant even see the 0.000 logs 2022-04-04 08:52:54 Hi folks! I'm trying to get FreeCAD working on my system. I'm using Sway and there's a bug in Coin3D that won't allow it to start on Wayland. I'm trying to build FreeCAD and Coin3D with a patch, but FreeCAD seems to depend on both ffmpeg4-dev and ffmpeg-dev (over qt5-qtwebengine-dev and vtk-dev), which are in conflict with each other. Do you have an idea on how to resolve this? 2022-04-04 09:13:19 hmmmm 2022-04-04 09:13:26 I have apk del -r mesa hanging for years 2022-04-04 10:06:47 Different question: After quite some time of searching the web without success, I just found out that the package "vlc" doesn't contain a GUI, but 'vlc-qt' does. Is this documented anywhere? If not so, shouldn't it be? 2022-04-04 10:21:22 Waweic: it's a common practice for alpinelinux to split things up 2022-04-04 10:23:35 I guess it could be documented somwhere, but I dont know where 2022-04-04 10:26:57 ikke: yeah, I can imagine usecases, where that would be really helpful. It's not bad, it's just different to the approach of most other distros 2022-04-04 13:57:28 anyway, I found that busybox adduser/addgroup have max id 256000, so I'll have to install "shadow" to be able to add a group with a higher gid 2022-04-04 13:58:24 why even limit max id's, other than if something would overflow, if it didn't use 32-bit ints 2022-04-04 14:07:32 jjakob: it was increased in Alpine from the Busybox default of 60000 in 2016: https://git.alpinelinux.org/aports/commit/main/busybox/busyboxconfig?id=6e956aff4700aadfd91db5b9df80ee8f79b2b90e 2022-04-04 14:08:57 Busybox has decided that it (i.e. adduser/addgroup) requires some arbitrary limit to be defined 2022-04-04 14:32:43 How did you exhaust this limit? 2022-04-04 14:39:53 long story, alpine is used in the docker images we use to run apps via docker-compose 2022-04-04 14:40:35 docker-compose secrets are bind mounts from the host, so in order for the process inside the container to be able to read the file, it needs to have read permission on the file 2022-04-04 14:41:19 so I add a "devgroup" with the same gid inside the container as the file gid on the host and set g+r on it 2022-04-04 14:41:51 the users on the host are FreeIPA domain users which default to the 790000000+ range 2022-04-04 14:42:47 yes the correct solution is to use docker swarm and stack deploy, yes I tried, https://github.com/docker/compose/issues/7773 2022-04-04 14:43:36 swarm secrets are not bind mounts, so you can configure them with any uid/gid/mode per service 2022-04-04 14:43:45 so this all becomes moot 2022-04-04 23:16:12 hi ! hit a error when trying to run xterm 2022-04-04 23:16:53 these are the lines after running xterm [ https://pastebin.com/Fx7PkYgW ] 2022-04-04 23:18:25 i'm not an expert in anything at all, but are you running xorg or wayland? cos normally that's the error i get when i'm running wayland and forget to use xwayland. 2022-04-04 23:20:51 @zote: we might not be running any of those ... how do i find out ? 2022-04-04 23:24:02 if you have a gui you're running one of them. you can do echo $XDG_SESSION_TYPE in a terminal if you have any working? 2022-04-04 23:24:36 but if you don't have a gui at all, and are just in a console, then xterm won't work at all because it's meant for a gui. 2022-04-04 23:25:16 is there anything necessary to make Alpine detect DisplayPort connections? 2022-04-04 23:25:38 eletrotupi: in what 2022-04-04 23:25:39 dmesg doesn't show anything whenever I connect my HDMI <-> DP adapter 2022-04-04 23:25:58 it is probably passive 2022-04-04 23:26:01 and it simply doesn't do anything at all, doesn't show on sway 2022-04-04 23:26:39 usually I would just plug the adapter and it would be ready (on Arch, so I figured it's prolly something missing on Alpine) 2022-04-04 23:26:40 have you tried running swaymsg -t get_outputs ? 2022-04-04 23:27:05 zote: yes, it simply shows the laptop display 2022-04-04 23:27:15 i have a DP - HDMI connection, and it just works, so i assume it's something installed by default. 2022-04-04 23:27:16 (this is the same build, but with alpine) 2022-04-04 23:27:26 hm, weird as fuck 2022-04-04 23:28:06 have you connected the display to the other end 2022-04-04 23:28:10 i wonder if it's because it's a laptop? mines a desktop so things may just be different. 2022-04-04 23:28:30 hm, maybe 2022-04-04 23:28:55 yes, it is the same monitor I use daily for work, I just switched to my personal desktop 2022-04-04 23:29:01 *personal laptop 2022-04-04 23:29:16 the weird is that I used the same adapter with arch and it just worked 2022-04-04 23:29:41 I'm mentioning Arch because I want to make it clear that this same layout/setup was indeed working properly two weeks ago 2022-04-04 23:29:43 try arch linux-lts 2022-04-04 23:30:19 it printed "x86_64" 2022-04-04 23:46:30 @zote: tΓ ... wanted to avoid adding full gui to stay cool on disk space 2022-04-04 23:53:17 yeah, xterm basically just gives you a terminal in the gui, so you don't really need it from the console. if you need more than one terminal application running at a time, tmux or screen work really well. 2022-04-04 23:56:47 ... was also searching for a package runinng a windowed sessions like terminal 2022-04-05 00:12:54 got to mention alpine WIKI seems really " easy " to read while picking up " things " - kudos to the authors. [ ... although, for US, this is still the worst time of the year for " new studies ", considering sinusitis and various allergies making US sleepy and tired in a matter of minutes. ] 2022-04-05 00:30:32 Hi, I just installed Alpine, and I am trying to install Sway as WM, following the guide in Alpine Wiki. When I run "# dbus-launch sway", it says there is an error in /etc/dbus-1/session.conf, line 1, column 0, no element found. I checked the .conf file with Nano, and it is empty. Below the error it says "EOF in dbus-launch reading address from bus daemon". 2022-04-05 00:30:32 What should I do? 2022-04-05 00:33:02 i assume it's not supposed to be empty 2022-04-05 00:33:44 after installing dbus in a new container, /etc/dbus-1/session.conf has a comment and then 2022-04-05 00:48:22 Thanks for your reply. I copied all that, but now the error says "column 155: mismatched tag" What could it be? Previously I tried using "apk fix dbus". 2022-04-05 00:50:24 Just in case, my version of debus is 1.12.20-r4 2022-04-05 01:25:42 you probably need to reset your configs 2022-04-05 09:38:48 Did anybody experience any issues with znc on alpine 3.15.* ? When I restart my znc, it fails to connect to one of my networks for some time. It says 'network unreachable', however irc server is reachable 2022-04-05 10:23:24 possibly DNS issues? 2022-04-05 10:33:36 znc connects to 2 servers (oftc and libera). This time it connected successfully to ofc and failed to connect to libera on the first try. Sometimes it fails to connect to oftc and connects to libera 2022-04-05 10:39:09 hmm 2022-04-05 13:03:08 I have a small mdev rule to change group of /dev/gpiochip[0-9] it works for the group but the mode is ignored 2022-04-05 13:03:12 gpiochip[0-9] root:input 0600 2022-04-05 13:03:14 that's my rule 2022-04-05 13:03:32 and the device is set to crwxrwxrwx 1 root input 254, 0 Jan 1 1970 /dev/gpiochip0 2022-04-05 17:57:16 hi all. i'm looking at the wiki page for bridge, Bridging for QEMU. the instructions says to echo stuff into /proc/sys/net/bridge, but i don't have that folder. the bridge package is installed, tap and tun modules are loaded. what am i missing? thank you 2022-04-05 18:36:59 opviel: try `modprobe br_netfilter` 2022-04-05 18:37:02 maybe the "bridge" kernel module 2022-04-05 19:16:33 thank you panekj , that was it 2022-04-05 20:19:24 Does anyone know if MUSL locale supports attribute bindtextdomain? And if not, is that something easily patched in the upstream project that is requesting that functionality? 2022-04-05 20:19:49 I have at least two APKBUILDs here stalled on that (Catfish for XFCE4 and Warpinator from Linux Mint for LAN file sharing) 2022-04-05 21:12:27 yeah, i can't get it to work. but then again, i also don't understand how it's supposed to work 2022-04-05 21:12:55 does anyone by any chance have alpine as a libvirt host, and the VMs on the same network as the host? 2022-04-05 21:13:27 it would seem that alpine only supports L3 bridges, yet there are instructions on passing through DHCP requests 2022-04-05 21:13:55 i've spent a couple days on this now and simply can't see my way forward 2022-04-05 21:15:36 if it's a L3 bridge, passing through DHCP requests would also mean confing the DHCP server for the different, internal VM network behind the bridge 2022-04-05 21:16:06 am i completely off the road? 2022-04-05 22:48:22 "L3 bridge" is a trash name, even moreso than OSI model to begin with. you could call it L2 bridge and L3 switch, but you shouldn't, because OSI model is wrong. accurate names would be Ethernet bridge and IP switch 2022-04-05 22:48:59 alpine supports exactly as much as the linux kernel supports, which is a wide range of switching and routing configurations 2022-04-05 22:49:10 s/switch/router/ 2022-04-05 22:49:10 Hello71 meant to say: alpine supports exactly as much as the linux kernel supports, which is a wide range of routering and routing configurations 2022-04-05 22:49:14 ... 2022-04-05 22:49:36 s/switch\>/router/ 2022-04-05 22:50:45 "L3 bridge" is a trash name, even moreso than OSI model to begin with. you could call it L2 bridge and L3 router, but you shouldn't, because OSI model is wrong. accurate names would be Ethernet bridge and IP router. alpine supports as much as the linux kernel supports, which is a wide range of switching and routing configurations 2022-04-06 03:53:54 Does anyone run XRDP server on Alpine? I am having a nightmare of a time hosting an XRDP server for some reason 2022-04-06 03:54:34 I'm not sure if I'm missing a dependency or something. It just doesn't do anything but allow clients to connect to a black screen. Missing user role in /etc/groups? 2022-04-06 06:01:35 not sure if necessary, but are these users in the video and input group? 2022-04-06 06:56:08 I'm sorry, I misspoke. Only just one user (local machine user), and yeah, they are in both input and video. 2022-04-06 06:56:30 I meant I tried to connect via multiple different RDP clients 2022-04-06 07:06:55 Dear Hello71. The term "L2 Bridge" is universally recognized throughout the industry, and furthermore both technically and syntactically correct as it is a virtual switch, backed only by one physical interface, using the bridge module, and indeed even bridge package. All of that said, if all you have to offer is sophistry and pedantry, I would ask you to kindly refrain from opening your mouth, so 2022-04-06 07:07:01 that I may get responses from people actually interested in a solution, rather than proving that they are Very Smart(tm). 2022-04-06 07:08:47 My question remains: how may I set up a L2 bridge (or virtual switch, if you prefer) on Alpine. The documentation all says the br0 interface gets an IP, which would make it a L3 bridge. 2022-04-06 07:09:35 The goal is to have libvirt-created VMs on the host's L2 network. I see references to passing through DHCP, but that doesn't make much sense for a L3 bridge. 2022-04-06 07:09:53 Actual help and/or a pointer in the right direction is still greatly appreciated. 2022-04-06 07:27:42 opviel: that doesn't make it a L3 bridge 2022-04-06 07:27:48 there's no such thing as an "L3 bridge" 2022-04-06 07:27:50 that'd be a router 2022-04-06 07:28:05 rather, the br0 interface represents both the bridge *and* the host connected to it (like the cpu port) 2022-04-06 07:28:36 it's always a L2 bridge; and you don't actually *need* to assign it an IP address if you don't want to – the address is to allow the host itself to participate in the subnet that the bridge creates 2022-04-06 07:29:12 grawity: this makes sense, this mighta been my mistake. i 2022-04-06 07:29:17 'll do some more testing and get back. thank you 2022-04-06 07:29:19 because if you put e.g. your physical eth0 in the bridge, then eth0 can no longer speak IP – it no longer works like a separate interface, but only as a bridge port 2022-04-06 07:29:38 so if the host itself still needs IP connectivity through the bridge, that's what the IP address on br0 is for 2022-04-06 07:30:35 grawity: but the internal bridge network should be the same as the host network, so a different IP for the br0 interface should not be necessary 2022-04-06 07:30:45 it is not a "different" IP 2022-04-06 07:30:52 it is the same IP moved to a different interface 2022-04-06 07:31:14 like I said br0 represents both the bridge *and* the host's connection to that bridge 2022-04-06 07:31:58 so i understand it right, once you have a br0 definition you do not need the eth0 definition anymore, in the /etc/network/interfaces = 2022-04-06 07:32:19 yes 2022-04-06 07:32:26 when looking at tshark, the dhcp queries from the vms seem to originate from that interface IP, and the responses don't show up. what am i missing? 2022-04-06 07:33:00 (cause they are being addressed to the host interface) 2022-04-06 07:33:56 gonna need to see the captures for that one 2022-04-06 07:35:25 grawity: on the way 2022-04-06 07:38:41 (also, what interface are you running the capture on? what interfaces are part of the bridge? is libvirt's networking adding some fancy ebtables rules?) 2022-04-06 07:40:29 grawity: https://www.toptal.com/developers/hastebin/dazahoquwi.yaml 2022-04-06 07:41:07 hmm, how sure are you that those requests come from your VMs, and aren't just renewals by the host itself 2022-04-06 07:41:18 what MAC address are they being sent from/to? 2022-04-06 07:41:34 grawity: very, i can log in to the console and run dhclient 2022-04-06 07:41:42 do they show up on `tshark -i br0`, as well as `tshark -i `? 2022-04-06 07:42:20 are eth0 and actually part of the bridge? (i.e. does `bridge link` say "eth0 ... master br0") 2022-04-06 07:42:38 is br0 itself up? (ip -br link ls br0) 2022-04-06 07:43:29 (the -br is for 'brief' not 'bridge', sorry if that got confsing) 2022-04-06 07:44:20 `bridge` is command-not-found even though the package is installed. 2022-04-06 07:44:25 i have the same dhcp requests when tsharking br0 2022-04-06 07:44:31 the bridge is state UP 2022-04-06 07:44:31 bridge would be part of iproute2 2022-04-06 07:44:41 I forget the busybox thing 2022-04-06 07:44:56 the interface definition for br0 has `bridge_ports eth0` 2022-04-06 07:45:26 yeah but actual state doesn't always reflect desired configuration 2022-04-06 07:45:32 `bridge link` shows eth0 and vnet1, which the vm is in 2022-04-06 07:46:09 ok, so 1) do the requests show up on a vnet1 capture? 2022-04-06 07:46:15 2) what source/destination MAC address do they have? 2022-04-06 07:47:42 1) they do. 2) still looking for the cli flags 2022-04-06 07:48:07 ip link show dev 2022-04-06 07:48:24 not sure tshark, maybe tshark -VO eth, or tcpdump -e, or termshark 2022-04-06 07:49:02 grawity: the requests are coming from the VM's MAC 2022-04-06 07:49:33 and they still have the weird source IP even on vnet1? 2022-04-06 07:50:35 grawity: on vnet1 we only have the dhcp discover 2022-04-06 07:50:51 on br0 we see the query actually going out to the dhcp server, from the bridge interface's IP 2022-04-06 07:51:19 the dhcp ack's coming back, to that iface address 2022-04-06 07:52:47 could it be that is trying to NAT even though the interface type is bridge? 2022-04-06 07:52:53 *libvirt is trying to 2022-04-06 07:53:11 it's not impossible 2022-04-06 07:53:40 though it would be bridge-level stuff (I remember libvirt tries to configure ebtables in some cases), not just regular iptables NAT 2022-04-06 07:54:23 i can assign a static IP to the VM, and that'll work through the bridge 2022-04-06 07:54:24 but I'm still waiting to hear what the packets actually look like 2022-04-06 07:54:39 it's just the dhcp requests that get lost 2022-04-06 07:54:47 can't really make even a good guess without that 2022-04-06 08:00:38 grawity: i've got packet captures here, but they don't give me much info. on vnet1 the dhcp discover goes out from 0.0.0.0 to 255.255.255.255, and on br0 it goes out from 10.22.0.128 to 10.22.0.1 2022-04-06 08:00:57 where would these ebtables rules be stored? 2022-04-06 08:02:09 no match for 'nat' within the vm's XML either 2022-04-06 08:02:38 ok so I'm not going to get an answer for those MAC addresses I guess 2022-04-06 08:02:52 what do you have in the libvirt network definition? (/etc/libvirt/qemu/network/default.xml or something) 2022-04-06 08:02:53 let me doublecheck 2022-04-06 08:03:00 do you have in there? 2022-04-06 08:03:15 I *think* mode="open" is what you generally need for libvirt to use a completely plain bridge 2022-04-06 08:04:04 see https://libvirt.org/formatnetwork.html (and somewhat related https://libvirt.org/formatnwfilter.html) 2022-04-06 08:04:34 it sounds like mode="bridge" has some filtering, mode="open" has none 2022-04-06 08:04:56 on vnet1 the mac is the VMs, on br0 it's the host's mac 2022-04-06 08:05:44 and just in case, check if there are any ebtables rules loaded 2022-04-06 08:06:54 ebtables has all ACCEPT, no special rules. i'll try with mode=open 2022-04-06 14:19:26 opviel: `sysctl net.ipv4.conf.all.bc_forwarding=1` (or in interfaces, `post-up sysctl net.ipv4.conf.$IFACE.bc_forwarding=1`) to allow broadcast forwarding through the bridge. Alternatively, if you're just trying to put VMs on your network, consider a macvlan. https://rentry.co/vi7qt LXD handles it trivially, and I know QEMU works with it too, so I presume libvirt would. 2022-04-06 14:20:51 opviel: This may be too basic for you, but is a decent overview of available virtual networking interfaces if you wish to explore another: https://developers.redhat.com/blog/2018/10/22/introduction-to-linux-interfaces-for-virtual-networking 2022-04-06 15:29:25 thank you nickersonm , i 2022-04-06 15:29:30 'll look into it some more 2022-04-06 15:34:15 hmm I don't recall bridges needing bc_forwarding 2022-04-06 15:34:22 they kind of already do that by default 2022-04-06 16:00:11 grawity: yes, good point. Looking at it again I don't know why I have that in my bridge setup. Perhaps I confused it with a NAT at the time. 2022-04-06 16:26:34 still struggling to figure it out. do i get it right that /etc/network/interfaces should not contain an eth0 definition, and instead have the ip on br0 with bridge_ports eth0 ? 2022-04-06 16:27:00 it's just the stupid dhcp... manually setting up an IP and netmask works lovely 2022-04-06 16:27:15 I'm not 100% sure about ifupdown specifics, but I think it should have an "iface eth0 inet manual" 2022-04-06 16:27:22 to still *bring it up*, only without an IP address 2022-04-06 16:31:55 yeah, turns out i was being a dumbass 2022-04-06 16:32:05 thank you grawity for taking so much time to help me 2022-04-06 19:55:33 Just getting started with alpine. Installed acf, shows as listening on 443, but can't connect. Can connect to 22. Is there a default firewall? Iptables shows accept. 2022-04-06 19:56:51 there is no default firewall 2022-04-06 19:57:18 are you sure it's listening on 0.0.0.0:443, if you're connecting from outside the machine 2022-04-06 19:57:23 some things just listen on localhost 2022-04-06 19:57:54 netstat -tln can give you some info 2022-04-06 19:57:59 ss shows it listening on *:443 2022-04-06 19:58:33 that shows :::443 2022-04-06 19:58:56 that's ipv6 only, if there is no corresponding 0.0.0.0 2022-04-06 19:59:03 but yeah, it's listening 2022-04-06 19:59:18 so either something in some *tables is actually blocking it, or it's failing to respond, etc 2022-04-06 19:59:24 what's the "can't connect" error 2022-04-06 19:59:42 can't connect == connection timeout 2022-04-06 20:00:07 Ncat: Connection timed out. 2022-04-06 20:01:08 Sounds like I need to figure how how to make it listen on ipv4. 2022-04-06 20:01:57 is acf acf-core 2022-04-06 20:02:21 (i don't actually even know what this is) 2022-04-06 20:02:35 but i'll take a look at it if you point me to what you ran 2022-04-06 20:02:54 setup-acf 2022-04-06 20:03:08 https://wiki.alpinelinux.org/wiki/Alpine_Configuration_Framework_Design 2022-04-06 20:05:01 hmm 2022-04-06 20:06:57 they all work for me 2022-04-06 20:07:12 connecting to https://localhost , 127.0.0.1, ip, hostname, anything 2022-04-06 20:07:18 in a browser 2022-04-06 20:07:26 didn't touch the config 2022-04-06 20:07:39 could you show `ip a` 2022-04-06 20:07:39 strange 2022-04-06 20:09:55 ip a > https://privatebin.net/?7b77ec408f22db8e#Gr8soPqmYET18fFc6KBwtTWmzkEznEfHVKvGAsQy8Za4 2022-04-06 20:10:42 hang on... 2022-04-06 20:12:16 Sorry, it was the hypervisor's firewall. 2022-04-06 20:12:33 I owe you a beer. 2022-04-06 20:14:41 :p 2022-04-06 20:14:45 also that only has a link-local ipv6 2022-04-06 20:15:48 I don't know much at all about ipv6. 2022-04-06 20:17:05 I really like this OS though. It's super fast. 2022-04-06 20:30:02 This is so cool. I am in nerd heaven. 2022-04-06 23:04:45 Alpine truly is a joy to use, isn't it? 2022-04-06 23:05:56 Incredibly niche question time: Is anyone using QEMU to host x86_64 OS with TPM/SecureBoot? I'm trying to hypervise Windows 11 within Alpine on hardware on the same machine that used to host Windows 11. Without TPM passthrough, I can't get Windows11 to boot under QEMU, I think. Unless I'm misconfiguring something else 2022-04-06 23:19:08 https://qemu.readthedocs.io/en/latest/specs/tpm.html 2022-04-06 23:19:15 perhaps something here stands out 2022-04-06 23:19:19 but there's a bunch of things 2022-04-06 23:19:51 slightly less verbose https://en.opensuse.org/Software_TPM_Emulator_For_QEMU 2022-04-06 23:20:03 we do have swtpm in testing 2022-04-06 23:25:10 isn't the point of tpm that you can't just pass it through 2022-04-06 23:26:07 otherwise you could decrypt bitlocker by just running it in qemu and then rummaging the memory 2022-04-06 23:44:04 I'm not sure how it would work. Maybe since the host isn't using the TPM the hypervized guest can? 2022-04-06 23:45:08 Thanks, psykose. I'll dig through that. I really need my Windows image working but I figured it was long since time to dogfood Alpine on my primary hardware and use it for everything 2022-04-06 23:45:21 Can't get ArcGIS working in Wine so... 😭 2022-04-06 23:46:05 I may have imprudently moved the timetable on this switch up too soon. I didn't trial, and I didn't research well enough since I ran into this issue. 2022-04-06 23:50:24 Saijin_Naib: do you NEED to have tpm, or could you just modify the image and bypass it? 2022-04-07 00:01:45 windows 11 needs it 2022-04-07 00:09:54 you can modify boot.wim to remove the checks, for all the requirements 2022-04-07 00:13:04 For an installed/running image captured with SysInternals disk2vhd? 2022-04-07 00:13:37 This was a physical on hardware installation I captured from the same hardware that is now hosting Alpine 2022-04-07 00:14:00 it's probably possible but just passing a qemu arg and installing something seems at least 450,123,532x less effort 2022-04-07 00:14:05 I'm not looking to install clean, and I know I can bypass checks for that by moving appraiserres 2022-04-07 00:14:20 ah that makes it a little harder. 2022-04-07 00:15:01 psykose: Not if I needed to keep my software environment. It's not just having Windows 11, I need all the software and configuration changes this image has. It needs to run as-is 😬 2022-04-07 00:15:29 yeah my point is moot if you cant rebuild the image. 2022-04-07 00:15:59 Ahhh, I did a stupid. Good to know. 2022-04-07 00:16:10 Have a link to that method anyway? I may eventually have no other option 2022-04-07 00:16:32 http://lambdacreate.com/posts/28 <- heres what I did to bypass it for my PXE server 2022-04-07 00:17:44 https://www.tomshardware.com/how-to/bypass-windows-11-tpm-requirement 2022-04-07 00:18:47 Beautiful, thank you so much 2022-04-07 00:20:27 NP hopefully it helps 2022-04-07 00:21:21 I will say that I have a w11 vm that doesnt use that hacky workaround, I just needed to add a TPM device in qemu and have it passthrough /dev/tpm0 2022-04-07 00:21:44 I used the CRB model for that. 2022-04-07 00:53:23 "I will say that I have a w11..." <- That's what I'm hoping to achieve here. And as a side-effect, get SecureBoot with self-signed stuff working for Alpine 2022-04-07 00:53:30 "I used the CRB model for that." <- CRB model? 2022-04-07 03:53:22 durrendal: Gotcha. Great tip! Works properly with a fresh Win10 install, so I think I'm getting closer to forcing my Win11 capture to work 2022-04-07 03:53:24 ACTION uploaded an image: (96KiB) < https://matrix.org/_matrix/media/r0/download/matrix.org/PWlnTvoCymmygJsIezvXoiHN/image.png > 2022-04-07 05:06:42 ACTION uploaded an image: (112KiB) < https://matrix.org/_matrix/media/r0/download/matrix.org/DYFjylQNEudEQOOmffQzAxCb/image.png > 2022-04-07 05:06:51 Brilliant! Clean install of Windows 11 works with your settings 2022-04-07 05:34:24 hey * 2022-04-07 05:34:39 anyone using Alpine as KVM/QEMU hypervisor? 2022-04-07 06:00:03 I am (if I understand your question) 2022-04-07 06:00:19 durrendal: You saved me. Days I've wasted on this haha 2022-04-07 06:07:59 Saijin_Naib: what do you use to create VMs, virt-manager? 2022-04-07 06:08:41 Yep! Keep in mind my perspective on this is as a novice. You'll need advice from more knowledgeable folks if that does not suit 2022-04-07 06:09:30 Above are screenshots of me running Win10 and Win11 in SecureBoot with TPM pass through within virt-manager 2022-04-07 06:25:22 Saijin_Naib: nice. I just started using alpine. Gonna check the wiki. Thanks 2022-04-07 09:48:36 PiRATA: virt-manager/libvirt is not needed, depending on your needs maybe you are better runing them directly, using custom scripts or qemu-openrc 2022-04-07 09:53:43 yeah, virt-manager needs X 2022-04-07 09:54:09 virsh and virt-install should be enough 2022-04-07 09:58:16 well, I mean that you can avoid whole libvirt, strictly speaking qemu-system should be enough :P 2022-04-07 10:18:20 true, gonna read qemu-system-x86_64(1). :) ty 2022-04-07 10:48:39 donoban: how do I know the number of vcpus available? 2022-04-07 10:49:13 do you mean in your host/desktop? 2022-04-07 10:49:41 yes, I have 4 phy cores 2022-04-07 10:49:55 you can check /proc/cpuinfo, top, etc.. 2022-04-07 10:50:09 nproc 2022-04-07 10:50:49 maybe you have theorically 8 vcpus if you have hyperthreding but I think that is not very good idea to take them in account 2022-04-07 10:51:11 it is fine for vms 2022-04-07 10:51:13 and most things 2022-04-07 10:51:26 if you don't want to use them then you should disable the smt and use the number of cores 2022-04-07 10:51:43 but if it's enabled it's better to pretend you have that many cpus, unless you are doing something specific 2022-04-07 10:52:15 well I didn't check peformance on VM's in some years but I have many experiences when too less was better 2022-04-07 10:52:32 you don't have to check vm performance, we already know the answer is 'very slow' 2022-04-07 10:52:33 :p 2022-04-07 10:53:07 now I remember one windows VM performing a huge task were 2 vcpus was clearly better than 4, I don't know if ht could affect 2022-04-07 10:53:13 multiple VMs can use the same vcpu? 2022-04-07 10:53:20 you can overcommit, yes 2022-04-07 10:53:22 windows is pretty bad handling multiple cores and intensive cpu process 2022-04-07 10:53:38 there's lots of cpu options 2022-04-07 12:38:18 donoban: but thats more regarding ACPI or something like that 2022-04-07 12:38:24 I have encountered that myself 2022-04-07 12:38:34 correct settings in the vm host would boost the performance raddically 2022-04-07 12:39:09 vcpu is just how many processes your vm guest will run at the vm host 2022-04-07 12:39:15 so you can overcommit as much as you like 2022-04-07 12:39:31 but most vm hosts limits your choice of num of vcpu's to the num of logical cores your vm host have 2022-04-07 12:39:45 but you can run multiple vm's with the same vcpu setting at once 2022-04-07 12:40:35 and if you use kvm/qemu dont forget to change cpu type into "host" unless you do clustering 2022-04-07 12:41:26 how do I see the console output from qemu? I'm using qemu-system-x86_64 with -display curses but stops outputing after the initial boot (alpine)? 2022-04-07 12:41:48 no idea, check on how proxmox does that =) 2022-04-07 12:41:55 they use kvm/qemu as backend 2022-04-07 12:42:23 and you can get the console there (even if its through the webbrowser but Im pretty sure it just pipes some qemu command output) 2022-04-07 12:42:34 -serial stdio as well 2022-04-07 12:42:52 and just -nographic instead of curses probably works 2022-04-07 12:43:16 nope, -nographic does not seem to work 2022-04-07 12:43:56 PiRATA: Probably your kernel command line options aren't sending console output to serial 2022-04-07 12:44:59 -append "console=ttyS0", or something like that 2022-04-07 12:45:03 depends how it's booting 2022-04-07 12:45:14 i think that only works if you pick a kernel directly actually 2022-04-07 12:45:19 >boot allows set tty comands? 2022-04-07 12:45:27 but he should get a getty with just -nographic 2022-04-07 12:45:44 PiRATA: could you paste your whole command? 2022-04-07 12:47:28 qemu-system-x86_64 -m 1024 -nic user -boot d -cdrom alpine-standard-3.15.4-x86_64.iso -hda alpine.qcow2 -display curses 2022-04-07 12:47:30 Apachez: sure, there are many aspects that affect. I just feel that too many vcpu can harm if no there is a workload that benefits from it 2022-04-07 12:49:02 uhM 2022-04-07 12:50:35 that also fails here, maybe you need to specefy a machine type / kvm accel etc.. 2022-04-07 12:50:53 try adding -machine type=q35,accel=kvm 2022-04-07 12:51:29 this: qemu-system-x86_64 -m 512 -boot c -nographic -cdrom alpine-virt-3.15.0-x86_64.iso -machine type=q35,accel=kvm 2022-04-07 12:51:31 works for me 2022-04-07 12:51:35 have to leave for a while, see you 2022-04-07 12:51:41 let me try 2022-04-07 12:51:56 ty donoban 2022-04-07 12:54:41 i did try to compile truecrypt on alpine-linux but it is not successful. run into minor error here and there 2022-04-07 12:54:54 anyone have a working truecrypt on alpinelinux platform ? 2022-04-07 12:54:57 for some reason -display gtk works, and none of the others do 2022-04-07 12:55:04 fsdfjekljjjdsadakl: what error do you get 2022-04-07 12:55:53 psykose: How much output is visible with the other display options? 2022-04-07 12:56:13 curses shows openrc then vanishes 2022-04-07 12:56:18 nographic default never shows anything after bios 2022-04-07 12:56:38 `qemu-system-x86_64 -m 4096 -nic user -boot d -cdrom alpine-standard-3.15.4-x86_64.iso -hda alpine.qcow2 -enable-kvm -machine type=q35,accel=kvm -device intel-iommu -cpu host -display gtk` works for me, with random stuff thrown in 2022-04-07 12:57:33 Hmm, showing openrc then vanishing is weird, so it's not some problem with the console parameter _or_ the initramfs. 2022-04-07 12:57:58 Maybe try `-serial stdio` or `-serial mon:stdio`? 2022-04-07 12:58:29 psykose: it's working now 2022-04-07 12:58:37 -serial stdio makes it start a vnc server, lol 2022-04-07 12:58:38 with -machine type=q35,accel=kvm 2022-04-07 12:59:04 psykose: Eh, that's weird. Don't think that's supposed to happen 2022-04-07 12:59:08 adding -nographic just does the same thing as before 2022-04-07 13:00:19 ty psykose 2022-04-07 13:00:52 donoban gave you that one 2022-04-07 13:17:44 yup, it's working properly. Trying OpenBSD atm and no issues so far :) 2022-04-07 13:53:52 accidentally deleted all /etc/init.d/ now time to not reboot 2022-04-07 13:55:27 apk fix \*-openrc should get them back 2022-04-07 13:55:41 aside from the few that are not from -openrc but the other one 2022-04-07 13:56:08 openrc, busybox-initscripts as well 2022-04-07 13:57:52 psykose, thanks! 2022-04-07 13:58:24 Maybe apk fix \* just to be sure that _everything_ is re-installed. 2022-04-07 13:58:52 that works for sure as well 2022-04-07 13:59:10 but i am 99% sure for almost every initscript it's just openrc busybox-initscripts \*-openrc 2022-04-07 13:59:14 there are a few stray ones that are not 2022-04-07 13:59:17 and are part of the main package 2022-04-07 13:59:50 generally though i've never broken anything with a fix \*, so it's fine 2022-04-07 14:17:49 anyone got the veracrypt working on alpinelinux? there was an error 'symbol not found error ' 2022-04-07 14:19:36 could you post the actual error 2022-04-07 14:23:22 pysykose , https://paste.linux.chat/?ba44e265eac6475a#6YHYB1X12g7kVayAj3GU59e6AyWwW58VfKbvzsA7Gjdf 2022-04-07 14:23:56 you installed something from edge while on 3.15 2022-04-07 14:24:46 pyskose, veracrypt available on edge only ..if i am not wrong 2022-04-07 14:25:12 you must upgrade your whole system to edge 2022-04-07 14:25:28 or better, just use cryptsetup 2022-04-07 14:26:30 Hello71, edge is in unstable stage ,right 2022-04-07 14:27:56 edge contains packages for the next stable release. it will necessarily have less testing 2022-04-07 14:28:15 whether it is "unstable" depends on your perspective and expectations 2022-04-07 14:28:56 truecrypt/veracrypt isn't particularly good to begin with so your "revealed standards" are low 2022-04-07 14:30:15 as we all know hello71 only makes the best of choices 2022-04-07 14:30:44 well, i try to make the best choices in my situation 2022-04-07 14:30:58 the best choices may be different in your situation 2022-04-07 14:31:18 A few small breakages during major updates aside (looking at you, OpenSSL and clang) Edge has been exceedingly stable for daily laptop usage 2022-04-07 14:32:06 Surely less problems than I had even between major Ubuntu upgrades (NON-LTS), and less breakage than Solus Shannon 2022-04-07 14:43:51 Saijin_Naib: what about steam and such? 2022-04-07 14:44:36 I use the Flatpak, and I've never had any breakage with the flatpak runtime that wasn't fixed by a reboot 2022-04-07 14:45:29 I wish we had it native for Alpine, but I doubt it is possible to get Valve to compile it for MUSL 2022-04-07 14:46:31 i'm inclined to say that steam could work with STEAM_RUNTIME=0 and/or gcompat 2022-04-07 14:46:32 And even then, I think it would not work right since Steam has 32bit and 64bit components and I don't think MUSL supports that yet 2022-04-07 14:47:45 hm, i forgot about multilib 2022-04-07 14:48:07 i think steam itself has a chance of working, it's just crappy electron 2022-04-07 14:48:12 but the games are probably hopeless 2022-04-07 14:49:02 since when steam is electron 2022-04-07 14:50:39 panekj yeah last I checked it was using the Source engine 2022-04-07 14:51:18 for games rather 2022-04-07 14:51:29 not for steam itself 2022-04-07 14:52:25 panekj: actually: In addition to the Steam client, several features were made available to developers, allowing them to take advantage of the cross-platform Source engine 2022-04-07 14:54:03 that was wikipedia's specific section for MacOS though 2022-04-07 14:54:11 so for Linux.. who knows 2022-04-07 14:55:21 the context of that is important 2022-04-07 14:55:33 "Steam for Mac OS X was originally planned for release in April 2010; but was pushed back to May 12, 2010, following a beta period. In addition to the Steam client, several features were made available to developers, allowing them to take advantage of the cross-platform Source engine, and platform and network capabilities using Steamworks" 2022-04-07 14:56:49 I doubt anyone would be insane enough to make a game launcher using game engine 2022-04-07 15:03:36 panekj; I play steam games on a weekly basis on my macs 2022-04-07 15:03:58 works just fine 2022-04-07 15:04:04 does two step autentication affect gmail/email clients (i have a cam that stopped sending emails)? 2022-04-07 15:04:40 frag: if your account has tfa turned on yes 2022-04-07 15:05:02 you have to setup your tfa for each client 2022-04-07 15:05:11 Tenkawa: you are mistaking what I'm talking about, I'm talking about Steam Client and what technologies it's made with, not what runs games 2022-04-07 15:06:34 if you have gmail 2fa you need to make a gmail app password to use for the devices to not need 2fa on them 2022-04-07 15:07:57 Tenkawa: hm! 2022-04-07 15:09:18 frag: I have several devices linked currently.. can be very annoying everytime you need to login 2022-04-07 15:09:32 my domain email stopped sending since they only allow normal email clients, so the cam was filtered out ... 2022-04-07 15:09:47 Tenkawa: indeed 2022-04-07 15:10:52 hi all, how do I go about getting nginx running on alpine? its installed but there is no init.d/nginx.. which nis what I was expecting. 2022-04-07 15:11:09 psykose, are you aware of web browser that do not take up too much memory on alpinelinux ? 2022-04-07 15:11:27 if they take up no memory they also usually can't really browse anything 2022-04-07 15:11:28 lynx? 2022-04-07 15:11:29 firefox and chromium take up too much memory on a single tab ~ 350mb ? 2022-04-07 15:11:32 w3m ? 2022-04-07 15:11:43 epiphany aka gnome web? 2022-04-07 15:12:14 Kingsy: that is in nginx-openrc and automatically pulled in with openrc installed, it sounds like you are running a container 2022-04-07 15:12:39 for simple people who do not know about internet. they just want to surf 2022-04-07 15:13:04 psykose: yeah I am running mcr.microsoft.com/dotnet/sdk:6.0-alpine. But its constructed with a DockerFile so I can just install what I need to. 2022-04-07 15:13:14 epiphany should work, it's *very* simple 2022-04-07 15:13:21 then you probably want to run `nginx -g 'daemon off;'` as the command 2022-04-07 15:13:22 so I need to install openrc ? 2022-04-07 15:13:40 no idea what you're trying to do 2022-04-07 15:13:53 I am trying to run nginx. 2022-04-07 15:14:02 ok -> nginx -g 'daemon off;' <- 2022-04-07 15:14:12 panekj, it is built using webkit. it will take up the same amount of memory like chromium ? 2022-04-07 15:14:19 whats the best way of managing it? supervisord ? 2022-04-07 15:14:29 doing nothing, you just run that 2022-04-07 15:14:50 well yeah. but I have more than one thing I need to run. one thing is the napp itself. which is the docker command. and the other now is nginx. 2022-04-07 15:14:57 then make two containers 2022-04-07 15:15:10 psykose: naaa I want them both in the same container. 2022-04-07 15:15:21 oftcguestoftc: although it's not the most lightweight engine, I don't expect it to be as demanding as chromium, you will need to try and find out 2022-04-07 15:15:44 Kingsy: use s6-overlay then 2022-04-07 15:16:08 although what you are trying to achieve is quite terrible 2022-04-07 15:16:28 panekj: I am lost.. why is running a nginx server and a dotnet application terrible? 2022-04-07 15:16:44 its like absolute basic stuff 2022-04-07 15:16:56 "I want them both in the same container." this part is terrible 2022-04-07 15:17:07 paneky, ok 2022-04-07 15:17:37 panekj: its terrible to have nginx and an app running in the same container on the same OS? I would say its just a basic setup. I don't see whats so bad. 2022-04-07 15:20:15 it seems to defeat the purpose of containers 2022-04-07 15:20:42 if it exposes an HTTP interface, there is no need for it to have a webserver running inside, because it goes against the idea that containers are not supervised, it unnecessarily complicates deployment 2022-04-07 15:22:20 it also is painful if you will be distributing that and someone uses own reverse proxy, e.g. traefik, caddy because now it's 2nd web server <-> nginx <-> app 2022-04-07 15:22:57 yes but the application I am using is dotnet. which given configuration is a pain to setup to listen on port 80. so a proxy to listen on 443 and 80 and hand off makes sense here. It means that other containers don't need ot know the funky port it listens on. 2022-04-07 15:23:32 you can link 2 containers together in own network, app + nginx 2022-04-07 15:23:53 kingsy, your nginx will setup to use proxy passthrough to read your dotnet application (eg. port number) 2022-04-07 15:24:12 fair enough 2022-04-07 15:24:23 dotnet application setup in port 9090 . proxy 80 to 9090 in your nginx config 2022-04-07 15:24:40 search around for the example.. lot of nginx tutorial example.. 2022-04-07 15:24:51 I mean, you can bind container to whatever port you want 2022-04-07 15:24:57 oftcguestoftc: I am fine with the nginx config. I just wanted to know nhow to install and run nginx 2022-04-07 15:25:00 anyway its all good. 2022-04-07 15:25:17 dotnet can listen on 666 and you can run container that binds that to 80 2022-04-07 15:25:23 on host 2022-04-07 15:25:31 install nginx on alpine --> apk add nginx 2022-04-07 15:26:04 I think they already know that 2022-04-07 15:26:10 panekj: yep but I want internal communication between containers within the network. which will be the internal port not the host exposed one. so without nginx that would be dotnets weird funky port. 2022-04-07 15:26:19 got to go 2022-04-07 15:27:29 Kingsy: if it's on inter-container network you can just tell nginx to go straight to dotnet 2022-04-07 15:27:48 but on inter-container network. iirc you can bind them to different port per container as well 2022-04-07 15:27:55 yeah I was going to just handle that using 127.0.0.1 within the same container. 2022-04-07 15:29:03 with 2 containers, you can tell nginx "here is the hostname with port, connect now" and set specific hostname for dotnet app container 2022-04-07 15:29:24 yeah ok, fair enough 2022-04-07 16:27:24 Speaking of Steam Flatpak, does anyone know if we can access the Steam flatpak directory on disk to sideload saves, etc from another Steam install? 2022-04-07 16:44:06 yeah, you can copy things into the flatpak install. i think it puts it all in ~/.var/app/com.valvesoftware.Steam 2022-04-07 16:48:14 it's just a normal steam directory in there 2022-04-07 16:48:19 i've moved saves/data around easily 2022-04-07 16:48:29 you can also find the wine/proton dirs in there and all that 2022-04-07 16:48:39 just like normal steam 2022-04-07 16:49:39 Oh wow, thank you both so much! 2022-04-07 16:49:54 First time working with it after using it on Windows for like... oh god... 17+ years now? 2022-04-07 16:50:13 Old πŸ‘΄ 2022-04-07 16:50:35 i've used linux/flatpak steam for long enough i've ran all sorts of weird stuff under it 2022-04-07 16:51:08 did you know if you have a game mod that has it's own executable, you can symlink it replacing the real game executable, and then proton will run it just fine :p 2022-04-07 16:51:25 you just press start game and it starts the correct thing from the ui 2022-04-07 16:51:28 Does yours actually register desktop files for installed games into your DE? Mine isn't... Wondering if I need an override in Flatseal or soemthing 2022-04-07 16:51:34 i never used that 2022-04-07 16:51:38 i always use the steam interface 2022-04-07 16:51:41 No, but that sounds interesting... Hmm 2022-04-07 16:51:43 so, no idea 2022-04-07 16:53:37 ACTION uploaded an image: (1818KiB) < https://matrix.org/_matrix/media/r0/download/matrix.org/nTqtZDofufgPOaRjSlKMMpnf/image.png > 2022-04-07 16:53:39 THIS IS AMAZING HAHA 2022-04-07 16:53:48 2022, year of Alpine Linux on my desktop 2022-04-07 17:09:56 that is certainly a game 2022-04-07 17:13:31 Indeed it is 😎 2022-04-07 17:13:48 One less stress about going Alpine on my desktop 2022-04-07 17:27:20 game it is 2022-04-07 17:49:12 just curious: why are man pages not included in base install? 2022-04-07 17:50:24 because they take space that most people won't be looking at 2022-04-07 17:50:36 if you want them they are a apk add man-pages/docs/whatever away 2022-04-07 17:50:47 yeah, already installed them 2022-04-07 17:51:17 or `apk add docs` to automatically install them for installed packages 2022-04-07 17:52:45 oh nice, gonna add it to the wiki 2022-04-07 17:59:27 i think it's already there 2022-04-07 18:05:19 didn't see it anywhere 2022-04-07 18:05:45 I added on https://wiki.alpinelinux.org/wiki/Post_installation 2022-04-07 18:06:53 oh 2022-04-07 18:12:40 neat 2022-04-07 18:16:37 apk is really fast 2022-04-07 18:16:44 Yeah, it is 2022-04-07 18:16:46 even faster than xbps 2022-04-07 18:17:05 For the intended usecase, it must be fast 2022-04-07 18:17:13 Same hardware installing Alpine and set of packages and updates is something like 1/5th the time (or less) of Debian-derived in my testing 2022-04-07 18:17:53 ime any setup was just network bound for me 2022-04-07 18:18:14 meanwhile on ubuntu i sat there as apt was extracting things for 8 minutes, with no network access 2022-04-07 18:18:45 apt just likes to do stuff 2022-04-07 18:19:52 i'm stuff 2022-04-07 18:20:00 i'm apt 2022-04-07 18:20:12 i'm like 2022-04-07 18:23:27 apk is also used for building and creating packages? 2022-04-07 18:23:45 currently no, but apk 3 will 2022-04-07 18:24:06 nice 2022-04-07 18:31:52 Oooh, that's a nice upgrade I didn't know about. 2022-04-07 18:37:00 keepassxc still on v2.6.6? Perhaps I can help. I updated the OpenBSD port 2022-04-07 18:40:58 2.7.0 uses botan2 instead of libgcrypt and libsodium 2022-04-07 18:41:31 where can I find more info on updating packages? 2022-04-07 18:41:49 wiki has information about creating/building packages 2022-04-07 18:42:56 found it 2022-04-07 18:42:57 ty 2022-04-07 18:43:30 to update, you would bump pkgver, set pkgrel to 0, build, test if it works, create MR to aports 2022-04-07 18:44:46 also update makedepends 2022-04-07 18:45:41 gonna play a bit with it. Thanks 2022-04-07 19:19:18 stupid question, but since OpenSSL is… well OpenSSL, and given than LibreSSL tends to break stuff, has https://sr.ht/~mcf/libtls-bearssl/ been considered for apk ? 2022-04-07 19:19:22 s/apk/ssl_client 2022-04-07 19:19:22 jvoisin meant to say: stupid question, but since OpenSSL is… well OpenSSL, and given than LibreSSL tends to break stuff, has https://sr.ht/~mcf/libtls-bearssl/ been considered for ssl_client ? 2022-04-07 19:28:22 ssl_client is openssl 2022-04-07 19:28:25 since today 2022-04-07 19:28:43 https://gitlab.alpinelinux.org/alpine/aports/-/commit/3b5e7c1fe308b1e2a1d028464f5d88a7969f10e6 2022-04-07 19:28:52 ah, you mean both 2022-04-07 19:28:54 hah 2022-04-07 19:28:59 i think it's preferred to make it openssl only 2022-04-07 19:29:06 so that there is only one crypto lib in the base image 2022-04-07 19:29:38 or more precisely, ncopa forgot to bump libretls after the last round of cves and had to make two releases 2022-04-07 19:29:43 and then spitecoded this :p 2022-04-07 19:30:06 indeed 2022-04-07 19:30:11 the question was more "BearSSL instead of OpenSSL, yay or nay?" 2022-04-07 19:30:22 i think nay for the same reason libretls was right there 2022-04-07 19:31:26 ack 2022-04-07 23:48:49 played a bit with abuild and sucessfully updated keepassxc (https://bpa.st/QFLQ) :) 2022-04-07 23:49:37 even though Jakub Jirutka updated before me to 2.7.1 2022-04-07 23:51:01 abuild is pretty cool. Way simpler than OpenBSD's tools :) 2022-04-07 23:59:16 https://git.alpinelinux.org/aports/tree/community/keepassxc/APKBUILD - Doesn't yubico keys require yubico-c-dev/ykpers-dev? 2022-04-07 23:59:39 probably 2022-04-08 00:00:05 it was removed from depends 2022-04-08 00:00:35 no, just pcsc 2022-04-08 00:00:39 according to cmake 2022-04-08 00:00:41 it's detected fine 2022-04-08 00:00:52 oh OK 2022-04-08 00:01:12 didn't catch that 2022-04-08 01:29:08 Do I need to do anything special to use abuild rootbld? 2022-04-08 01:29:17 I'm already a member of abuild group 2022-04-08 01:29:34 did you install abuild-rootbld? 2022-04-08 01:30:09 Yep. Do I need to logout or reboot for any reason after that? 2022-04-08 01:30:36 I don't see why would you need to do that 2022-04-08 01:32:56 Good, me either. Do I need any other group membership? 2022-04-08 01:33:04 I'm wheel and use doas 2022-04-08 01:33:20 I know chroot needs root, right? 2022-04-08 01:33:25 no 2022-04-08 01:33:36 it runs under your account 2022-04-08 01:33:51 it's bwrap 2022-04-08 01:34:07 chroot on steroids 2022-04-08 01:34:26 Oooh, new stuff to look up 2022-04-08 01:51:11 any error Saijin_Naib? 2022-04-08 01:51:27 i was able to run abuild rootbld 2022-04-08 02:01:53 Yes, sorry, but no longer at my desktop. Complains about .git not being a directory 2022-04-08 02:37:35 ACTION sent a code block: https://matrix.org/_matrix/media/r0/download/matrix.org/oMOmHBKwUciEfkUCyQJEbUQl 2022-04-08 02:38:53 why is username there twice? 2022-04-08 02:39:10 ahm, your repo, never mind 2022-04-08 02:40:07 Saijin_Naib: is that aports repo? 2022-04-08 02:52:36 Nah, just a flat folder with a single APKBUILD. Testing some changes to redshift 2022-04-08 02:54:17 you need aports repo 2022-04-08 03:37:32 Ohh, okay 2022-04-08 03:37:47 Thank you! 2022-04-08 08:19:04 so basically vim has a new release every git commit 2022-04-08 10:06:21 Hello there, question, how can I paste in a code block as fancy way @Saijin_Naib did? 2022-04-08 10:09:34 gopher22678381: If you're on Matrix, with three backticks (```) before and after the stuff you want in a block. 2022-04-08 10:09:36 you can use a pastebin 2022-04-08 10:09:55 that feature specifically is a matrix->irc bridge thing, where it converts any multiline message into an upload for irc side 2022-04-08 10:10:08 since irc only has one line 2022-04-08 10:10:33 psykose: In this case Saijin_Naib did a code block, not a multiline message 2022-04-08 10:11:16 it's the same thing, the code block is multiline. the bridge calls them different but it's the same concept 2022-04-08 10:11:20 ..pretty sure anyway 2022-04-08 10:11:31 ktprograms psykose thank you, I don't use matrix. pastebin it is then 2022-04-08 10:11:42 psykose: Only difference is that it's rendered as a code block in Matrix 2022-04-08 10:11:49 yeah, but you can mix the two can't you 2022-04-08 10:11:58 ```some stuff\n more stuff```\n message 2022-04-08 10:12:02 I'm using thelounge, and I see that it actually previews the contents if I paste it in 2022-04-08 10:12:16 Think so, not sure how that's displayed by the bridge 2022-04-08 10:12:25 just nitpicking anyway so... 2022-04-08 10:12:27 just a link like always 2022-04-08 10:13:28 I meant what text is put before that. When it's a multiline message the first line is sent and the rest is in the link. (that's the one annoying part about the Matrix <-> IRC bridge) 2022-04-08 10:13:47 heh, i also find that annoying 2022-04-08 10:13:56 I see them like this: https://i.imgur.com/lFJ7grk.png 2022-04-08 10:14:15 actually this link to the image is previewed as well 2022-04-08 10:15:00 yeah, thelounge detects the matrix links and displays them inline 2022-04-08 10:15:04 and a lot of other links 2022-04-08 10:19:07 ok, so with that out of the way, I have a question, specially for networking/ipv6 gurus. I have a native ipv6 config (or dual-stack in other terms) from my ISP. notied that in Alpine, by default, if I set up an inet6 static (local, eg. 2001:fdda::1/64) address for an interface, I also get the RA address (external, unique, SLAAC, DHCPv6 forwarded) 2022-04-08 10:19:07 from my router. Which is good and this is how I intend it to be. 2022-04-08 10:20:36 however, this ins not the case on other distris, like debian, by default. there, when there is a static inet6 config, the RA address is not bind to that same interface. what options (referring to /proc/sys/net/ipv6/conf/...) does Alpine use by default for this behaviour? 2022-04-08 10:22:23 of course, in case if I configure inet6 auto (/etc/network/interfaces) in debian, I still get the (RA, SLAAC + local ipv6 address assigned any way), but not the custom static v6 of course 2022-04-08 10:27:03 that sounds like ifupdown does something extra with the static 2022-04-08 10:34:55 there's no other ifupdown script in Alpine by default really, apart from the DAD scrip (duplicte address detection, which is located here /etc/network/if-up.d/dad) 2022-04-08 10:35:19 no idea tbh 2022-04-08 10:36:20 isn't that a matter of sysctl settings? 2022-04-08 10:36:47 could be that most directives under /proc/sys/net/ipv6/conf/all are enabled (or 1) by default in Alpine? which is not the case under Debian? 2022-04-08 10:37:18 how can I cat * all the files and show the filename in stdout? 2022-04-08 10:37:19 dunno, compare 2022-04-08 10:37:20 https://img.ayaya.dev/jfImZeVFLRi9 2022-04-08 10:37:28 gopher22: grep can do that 2022-04-08 10:37:35 sysctl just prints them all by default 2022-04-08 10:38:02 grep -r . /proc/sys/net/ipv6/conf/all/ 2022-04-08 10:38:04 right 2022-04-08 10:38:43 ok, we're onto something here 2022-04-08 10:39:04 sysctl -a -r net.ipv6.conf.all 2022-04-08 10:39:21 even better 2022-04-08 10:39:39 the one i posted has exactly the same output smh 2022-04-08 10:39:57 oh 2022-04-08 10:42:41 net.ipv6.conf.all.accept_ra_pinfo 2022-04-08 10:42:46 net.ipv6.conf.all.autoconf 2022-04-08 10:43:09 net.ipv6.conf.all.accept_ra 2022-04-08 10:43:16 those seems to have influence on that 2022-04-08 10:43:56 yes, and those are enabled by default in Alpine as it seems 2022-04-08 10:45:02 Or not explicitly disabled 2022-04-08 10:46:02 the kconfig settings don't seem to be related to those 2022-04-08 10:46:06 so i think those are just the default 2022-04-08 10:46:19 and the sysctl.conf files aren't filled with anything by default i don't think 2022-04-08 10:46:23 I think I found the 'issue' 2022-04-08 10:46:27 just have to confirm 2022-04-08 10:47:14 yes, sysctl.conf is empty by default 2022-04-08 10:50:52 Are alpineconf 2021 recordings available? 2022-04-08 10:57:44 Ermine: There's a link to it on https://alpinelinux.org/conf/, but it seems like the site's certificate expired yesterday. 2022-04-08 10:58:29 Ariadne: You set up that site, right? 2022-04-08 10:58:45 Moreover, it returns 502 2022-04-08 10:59:06 wfm 2022-04-08 10:59:10 works for me 2022-04-08 10:59:13 oh, the link 2022-04-08 10:59:23 ah 2022-04-08 10:59:29 bbb.dereferenced.org 2022-04-08 10:59:34 yea 2022-04-08 10:59:34 that's managed by ariadne, indeed 2022-04-08 12:03:30 ok, for my "issue" a simple fix: add autconf 1 and accept_ra 2 into interfaces config for the respective inet6 stanza 2022-04-08 12:04:28 seems like Alpine does this imlicitly and obeys to the systclt rules, whereas on other distros, some scripts mock with sysctl stanzas 2022-04-08 12:13:59 We don't set any sysctl config, so what happens on alpine is default linux behaviour 2022-04-08 12:14:38 It's the kernel that takes care of these things 2022-04-08 13:12:10 bbb.dereferenced.org is down unfortunately 2022-04-08 13:12:37 what if we made it point upwards 2022-04-08 13:20:10 Was that part of the server that did not survive? 2022-04-08 15:22:08 hi, anyone knows a simple/minimal calendar/todo/remminder that sends dbus notifications? I tested with at¬ify-send but I found it pretty messy 2022-04-08 15:53:56 I think Orage has dbus support... I have it packaged in testing 2022-04-08 16:09:06 thanks Saijin_Naib , I think that I will try to hack a little on todo.txt 2022-04-08 16:10:47 or just create an script that gets the task name from 'todo list | dmenu', a time offset from now and calls at... 2022-04-08 17:46:44 is there a easy way to install all relevant -lang packages for already installed packages? 2022-04-08 17:47:11 apk add lang ? 2022-04-08 17:53:26 that was unexpectedly easy 2022-04-08 17:53:32 thank you 2022-04-08 18:17:30 I have a system with a usb disk I need mounted at boot time. Right now it seems to be available after `localmount`. kernel cmdline params already have `modules=usb-storage,sd-mod,uas` but the disk shows up 15s after boot.. any suggestions ? 2022-04-08 20:54:03 anyone using obs on 3.15? for me it sigsevs without saying much why 2022-04-08 20:55:02 nothing in backtrace either? 2022-04-08 21:00:33 mhm, with verbose it says that gpu not supported 2022-04-08 21:00:43 what is the gpu 2022-04-08 21:00:58 i don't think it needs any gpu support just to start.. 2022-04-08 21:01:03 amd rx590 2022-04-08 21:02:31 https://usercontent.irccloud-cdn.com/file/mbpT4etj/image.png 2022-04-08 21:02:45 https://tpaste.us/jPNm 2022-04-08 21:03:13 that doesn't look like anything segfaults 2022-04-08 21:03:40 fish: Job 1, 'obs' terminated by signal SIGSEGV (Address boundary error) 2022-04-08 21:03:43 without --verbose 2022-04-08 21:04:00 so --verbose makes it not segfault? 2022-04-08 21:04:30 apaprently 2022-04-08 21:04:41 s/apaprently/apparently 2022-04-08 21:04:41 panekj meant to say: apparently 2022-04-08 21:04:43 what is the backtrace 2022-04-08 21:05:21 how I can get it 2022-04-08 21:05:28 gdb obs 2022-04-08 21:05:29 bt full 2022-04-08 21:05:55 `run` in between those two 2022-04-08 21:06:36 No stack. 2022-04-08 21:07:13 it segfaults and then there is no stack 2022-04-08 21:07:16 something is very broken then 2022-04-08 21:07:37 do take a picture though because i don't trust you 2022-04-08 21:08:02 smh 2022-04-08 21:08:07 :p 2022-04-08 21:08:09 https://usercontent.irccloud-cdn.com/file/0HcAPjGR/image.png 2022-04-08 21:08:25 that has no segfault 2022-04-08 21:08:28 it exited normally 2022-04-08 21:10:22 but aside from this, idk 2022-04-08 21:10:34 C: 2022-04-08 21:10:35 do you have mesa-dri-gallium installed, etc etc 2022-04-08 21:10:42 can you run.. random other gpu stuff 2022-04-08 21:12:07 what "random other gpu stuff" could it be 2022-04-08 21:13:11 no clue 2022-04-08 21:13:15 eglinfo output, glxgears 2022-04-08 21:13:50 https://usercontent.irccloud-cdn.com/file/VZ1sK6oc/image.png 2022-04-08 21:14:56 amdgpu_device_initialize: amdgpu_query_info(ACCEL_WORKING) failed (-13) 2022-04-08 21:15:26 that doesn't mean anything because there are lots of things in there 2022-04-08 21:15:33 that specifically looks like gbm which will fail without root 2022-04-08 21:15:46 stop trimming things smh 2022-04-08 21:16:20 https://tpaste.us/Bkgb 2022-04-08 21:17:04 ah 2022-04-08 21:18:13 you can trace it a bit more, like in https://gitlab.freedesktop.org/mesa/drm/-/issues/52 2022-04-08 21:19:09 libgl_debug and some strace 2022-04-08 21:49:59 panekj: does it work in xorg? 2022-04-08 21:54:42 no idea 2022-04-08 22:00:18 Speaking of inspecting crashes... Does this mean xfce4-notifyd needs a rebuild? 2022-04-08 22:00:19 [ 485.600224] xfce4-notifyd[5610]: segfault at 7ff4ed86f0b3 ip 0000561eb4412c34 sp 00007ffd0c524e90 error 4 in xfce4-notifyd[561eb440b000+a000] 2022-04-08 22:00:19 [55332.815248] xfce4-notifyd[3742]: segfault at 7f45f927662d ip 00007f45fb4d2836 sp 00007ffc739256e8 error 4 in ld-musl-x86_64.so.1[7f45fb497000+48000] 2022-04-08 22:04:48 probably not 2022-04-08 22:05:48 Any particular direction I should be exploring? It works as root for some reason 2022-04-08 22:38:30 did you turn it off and on again? ;) 2022-04-08 22:39:02 does notifyd sefault each attempt or only 14 hours in between? 2022-04-08 22:44:32 Looks like after suspend/resume 2022-04-08 22:44:35 Upstream bug report time? 2022-04-08 22:44:52 We are a minor release behind... and the patchnotes indicate big memory management fixes 2022-04-08 22:44:57 Could that likely be a lucky fix? 2022-04-09 08:11:55 Hey, what's supposed to be the correct way to load a kernel module at startup on Alpine Linux? I've already tried registering the module inside of /etc/modules-load.d/ and /etc/modules, I've also executed rc-update add modules but nothing worked so far - the module is not loaded when I reboot. 2022-04-09 08:15:54 What I'd like to automate is 'sudo modprobe uhid'. 2022-04-09 08:40:58 rokf: adding a file ending with .conf in /etc/modules-load.d/ should work 2022-04-09 09:47:53 note it needs the .conf 2022-04-09 09:48:21 /usr/lib/modules-load.d/*.conf; do 2022-04-09 09:48:21 for f in /lib/modules-load.d/*.conf \ 2022-04-09 10:47:45 ikke: I've tried that location and it didn't work. I'll try the other two that psykose suggested 2022-04-09 10:50:25 For anyone else following along, those module folders are used / listed inside of the '/etc/init.d/modules' script. 2022-04-09 10:55:29 it checks that one too 2022-04-09 10:56:36 the /etc/modules-load.d/*.conf takes precedence 2022-04-09 10:57:00 you can edit stuff, then manually start `modules` to see if the things you add get loaded 2022-04-09 11:00:17 Hmm, neither the '/lib' and '/usr/lib' location isn't picked up :) Seems like the '/etc/init.d/modules' doesn't run at all. I'll try to figure out what's happening. Suggestions welcome :D 2022-04-09 11:00:33 `service modules start` 2022-04-09 11:00:34 what happens 2022-04-09 11:01:08 It says 'WARNING: modules has already been started'. 2022-04-09 11:01:14 ok 2022-04-09 11:01:15 now stop it 2022-04-09 11:01:17 and start it again 2022-04-09 11:07:34 What is the module in question? 2022-04-09 11:13:01 Btw you can remove -q flag in modprobe invocations in /etc/init.d/modules 2022-04-09 11:32:04 Are there any reasons why setup-apkrepos sets repositories to be accessed with http and not https? 2022-04-09 11:33:18 psykose: I've done that now. Ermine: The module in question is uhid. I've now removed the -q flag from the modprobe invocations, thanks for that tip :) I'll try restarting the computer again now. 2022-04-09 11:33:40 done what now? 2022-04-09 11:33:42 what did it say 2022-04-09 11:33:48 did the modules you added get loaded 2022-04-09 11:33:48 etc 2022-04-09 11:39:24 psykose: The first time I ran stop it just gave me a warning that I'm stopping a sysinit service. Start just said that it's loading modules, nothing else. The second time I ran stop it also mentioned that it's caching service dependencies. When I execute 'rc-update' it says that 'modules' runs at boot in the 'default' and 'sysinit' runlevels. 2022-04-09 11:39:46 yes, and did the modules get loaded or not 2022-04-09 11:39:50 you can see in `lsmod` 2022-04-09 11:40:14 it should not be running in sysinit 2022-04-09 11:41:31 I've added that 'sysinit' part myself, it only had 'default' out of the box when I've added it with 'sudo rc-update add modules' for the first time. I can remove that. 2022-04-09 11:45:49 I can't really tell if the 'uhid' module got reloaded with the modules service restart since I've previously loaded it manually through 'modprobe'. I can try removing it and then I do the 'service modules stop/start' again. 2022-04-09 11:48:21 Rebooting, BRB. 2022-04-09 11:52:51 yea, you can just `rmmod` things and then restart the service 2022-04-09 11:53:37 The service restart didn't help for some reason, the modules weren't picked up. 2022-04-09 11:54:49 did modprobe output anything? 2022-04-09 11:55:06 can you `ls -l /etc/modules-load.d` and `cat /etc/modules-load.d/*` 2022-04-09 11:57:56 Ermine: Is there a log file that's supposed to contain that output? 2022-04-09 11:58:13 running modprobe yourself reports something if it fails 2022-04-09 11:58:19 It should appear on tty 2022-04-09 12:00:19 psykose: 'sudo modprobe uhid' works and doesn't return any errors. The keyboard and mouse work after I execute that command. That's the reason I'm trying to run the uhid module. 2022-04-09 12:00:30 can you post the output of those two things i posted 2022-04-09 12:04:21 psykose: The folder where I currently have the single file is '/lib/modules-load.d/', it should work with multiple folders according to the script that runs the modprobe command on the files. The outputs are '-rw-r--r-- 1 root root 4 Apr 9 12:40 uhid.conf' and 'uhid%'. The file is named 'uhid.conf'. 2022-04-09 12:05:01 what is in the file 2022-04-09 12:06:12 Just the word 'uhid'. 2022-04-09 12:06:48 delete the uhid% file, move the .conf to /etc/modules-load.d, rmmod uhid, then restart the service 2022-04-09 12:21:31 psykose: It works now after a reboot. Either the runlevel in rc-update, a missing newline in the `.conf` file or the `service modules stop/start` must've fixed it, because I've already had the file in the correct folder before and it wasn't picked up. 2022-04-09 12:21:59 Thanks for the help! :) 2022-04-09 12:23:46 It could also be of some kind of side-effect of rmmod -f (I couldn't remove it without the force flag because the uhid module was in use). 2022-04-09 12:24:54 initially it would be the runlevel - the load would fail at sysinit since that is before modules are available i think 2022-04-09 12:25:01 or maybe not, fuck if i know what the runlevels really do :p 2022-04-09 12:25:12 stray lines don't matter 2022-04-09 15:36:22 Which TERM should I set on serial console? 2022-04-09 15:36:51 depends on how you connect to it 2022-04-09 15:36:54 The machine runs in qemu and I access it via virt-manager 2022-04-09 15:38:47 that just forwards to your local terminal no? 2022-04-09 15:39:04 it appears in virt-manager window 2022-04-09 15:39:27 i guess something like xterm should work fine 2022-04-09 15:39:39 usually people dont want 256 color fu on a serial line :p 2022-04-09 15:40:25 The problem is that vim does not work correctly (and vi does) 2022-04-09 15:40:55 try vt100 maybe? 2022-04-09 15:41:24 or use virsh console on a real terminal emulator :p 2022-04-09 15:42:19 Oh wow, thank you! 2022-04-09 15:42:27 yw :) 2022-04-09 16:33:52 Can confirm K8s works on 3.15.4 2022-04-09 18:14:54 is there a way to setup without the iso ? 2022-04-09 18:15:08 something like debootstrap ... 2022-04-09 18:16:53 something like alpine-conf 2022-04-09 18:17:06 genericum: this page links some projects that can install alpine in a chroot: https://wiki.alpinelinux.org/wiki/Chroot 2022-04-09 18:17:52 just the usual setup-* scripts 2022-04-09 18:18:53 Ermine, perfect thanks 2022-04-09 20:20:25 how can I display the list of installed packages ? 2022-04-09 20:20:38 apk list -I 2022-04-09 20:21:31 and where are all the files for the apk repos ? 2022-04-09 20:21:45 like packages? 2022-04-09 20:21:55 yup 2022-04-09 20:22:02 scripts 2022-04-09 20:22:12 or c programs actually 2022-04-09 20:22:24 You've made it even more confusing 2022-04-09 20:22:37 are you looking for builds scripts or package files? 2022-04-09 20:22:37 i thought the build system uses c 2022-04-09 20:22:45 build scripts 2022-04-09 20:22:51 build system uses whatever package requires to build 2022-04-09 20:23:10 must be in some language too ... 2022-04-09 20:23:19 eager to find out 2022-04-09 20:24:00 havent seen any yet 2022-04-09 20:24:15 alpine uses posix shell mostly for package build script definition 2022-04-09 20:24:33 builds scripts are in https://gitlab.alpinelinux.org/alpine/aports repo 2022-04-09 20:25:39 if I download a package its a bin ? 2022-04-09 20:26:09 like downloading *.apk file or what exactly? 2022-04-09 20:26:18 *.apk is tar archive 2022-04-09 20:26:23 apk add foo 2022-04-09 20:26:44 `apk add` downloads an .apk file and unpacks it 2022-04-09 20:27:11 if I were to simply describe it 2022-04-09 20:27:30 ok now that I know that its a tar ... 2022-04-09 20:28:03 one more problem my x server doesnt start :( 2022-04-09 20:28:18 check its log 2022-04-09 20:28:46 cat /var/log/Xorg.log | less 2022-04-09 20:29:08 seg fault 2022-04-09 20:31:07 ok deleted vesa module 2022-04-09 20:31:25 now i got "devices detected but none match those in config file" 2022-04-09 20:36:47 hope i got it right now ... lets reboot laters ... 2022-04-09 20:55:57 i don't think deleting random things helps 2022-04-09 20:57:38 but rm -rf /* helped me 2022-04-09 20:59:20 FIX your ALPINE INSTALL with this ONE WEIRD TRICK they don't WANT YOU TO KNOW 2022-04-09 21:05:33 something something forbidden knowledge 2022-04-09 21:05:45 "FIX your ALPINE INSTALL with..." <- :D 2022-04-09 21:05:56 ;3 2022-04-09 21:08:15 is there any display manager that doesn't depends on xorg 2022-04-09 21:11:26 gdm 2022-04-09 21:11:34 in theory, i think we pull in xorg anyway 2022-04-09 21:11:50 it pulls xorg 2022-04-09 21:12:05 greetd 2022-04-09 21:12:18 though have fun configuring that stuff 2022-04-09 21:12:26 I tried it once 2022-04-09 21:12:43 with various greetd backends, all suck 2022-04-09 21:18:25 psykose: Maintainers HATE HIM 2022-04-09 21:18:36 hate who 2022-04-09 21:19:43 Ads about such tricks contain this part 2022-04-09 21:19:58 hate me 2022-04-09 21:19:59 ah 2022-04-09 21:20:04 right yes 2022-04-09 22:16:54 Where can I find the source for the website? I noticed that the columns in the "Status of Alpine Linux mirrors" tables aren't ordered correctly 2022-04-09 22:16:58 https://mirrors.alpinelinux.org/ 2022-04-09 22:18:59 i've no idea where mirrors is, but most of the website stuff is https://gitlab.alpinelinux.org/alpine/infra/alpine-mksite/ 2022-04-09 22:21:48 it's https://gitlab.alpinelinux.org/alpine/infra/mirror-status 2022-04-09 22:22:29 Thanks! 2022-04-09 22:22:41 ordering is correct is though 2022-04-09 22:23:28 https://img.ayaya.dev/3C3AngN1iD0q.png 2022-04-09 22:23:32 For me the columns go: v3.10 v3.11 v3.12 v3.13 v3.14 v3.15 v3.5 v3.6 v3.7 v3.8 v3.9 2022-04-09 22:23:39 they are indeed wrong 2022-04-09 22:23:54 5 > 15 2022-04-09 22:23:56 https://gitlab.alpinelinux.org/alpine/infra/mirror-status/-/merge_requests/3#note_212173= 2022-04-09 22:24:03 er 2022-04-09 22:24:09 drop the note, but there is already an mr for this 2022-04-09 22:24:13 as expected, it will never be merged 2022-04-09 22:24:56 there's even a second one https://gitlab.alpinelinux.org/alpine/infra/mirror-status/-/merge_requests/5 2022-04-09 22:25:19 some day 2022-04-09 22:26:20 "they are indeed wrong" only if you expect it to be sorted by minor version 2022-04-09 22:27:12 they are not wrong if you are really stupid, yes 2022-04-09 22:27:41 Can I add https links? 2022-04-09 22:27:46 to what 2022-04-09 22:27:50 where 2022-04-09 22:28:10 who? 2022-04-09 22:28:50 mirror.yandex.ru line for example, but other mirrors can be checked for https support 2022-04-09 22:29:00 those are the links the mirrors give 2022-04-09 22:29:21 The second MR for this seems more promising - can we poke someone about this? 2022-04-09 22:29:27 if they didn't told us that they provide https, then they aren't presented with https support 2022-04-09 22:29:30 ikke: just merge it already ^ 2022-04-09 22:29:46 Ah ok 2022-04-09 22:30:03 i can add it anyway, but i don't see much point 2022-04-10 05:20:30 psykose: 409 2022-04-10 05:21:31 ikke: 426 2022-04-10 05:22:27 429 2022-04-10 05:23:09 303 2022-04-10 05:25:02 ( i did mean your own one ) 2022-04-10 05:25:04 also goodnight 2022-04-10 05:32:54 o/ 2022-04-10 11:45:40 Hello, I tried to connect to irc.adelielinux.org via ssl with znc, but znc doesn't trust its certificate for some reason 2022-04-10 11:46:55 Ermine: the correct server address with SSL support is irc.interlinked.me. irc.adelielinux.org is just a CNAME alias to that name, but the certificate there doesn't include irc.adelielinux.org as one of its subject alternate names 2022-04-10 12:00:04 Ok, got it 2022-04-10 12:56:16 raven523: so when will it be fixed? := 2022-04-10 12:56:18 :) 2022-04-10 12:56:51 probably never because of how the certs for *.interlinked.me are set up 2022-04-10 12:57:44 I could see about getting one for irc.interlinked.me and irc.adelielinux.org but that will require coordinating with a lot of people and probably messing with CAs 2022-04-10 12:59:01 At least this proble is not related to alpine 2022-04-10 13:00:16 #adelie on irc.interlinked.me is the official channel, but interlinked.me is run by a separate organization 2022-04-10 14:58:08 hm, I just installed `gcc`, but it cant find Srcrt1.o, crti.o 2022-04-10 14:58:16 (just doing gcc hello.c) 2022-04-10 14:58:46 is there something I'm missing? 2022-04-10 14:59:38 oh, I miht need musl-dev 2022-04-10 15:29:46 Is it possible to register my own hook to run after some package is updated? In my case, I would like to run a shell script every time linux-lts is updated. Is that possible with apk? 2022-04-10 15:31:29 make a package with the desired trigger 2022-04-10 15:36:05 you can look at the mkinitfs trigger for an example 2022-04-10 15:37:28 I'm trying to understand it right now... triggers="$pkgname.trigger=/usr/share/kernel/*", ok, that looks fine, one detail I'm missing is where/how do I specify the script to run. 2022-04-10 15:38:51 $pkgname.trigger 2022-04-10 15:39:58 Oh, so when the documentation say `scriptname=pathlist where scriptname is the (sub)package name`, it actually means that script or $pkgname is executed and there is no way to override it? (Just checking if I got it right.) 2022-04-10 15:40:46 *script named $pkgname 2022-04-10 15:44:12 what? 2022-04-10 15:50:47 Well the documentation says that it `scriptname' is the `(sub)package name'. So if I have create a package linux-lts-foo, that contains /usr/bin/linux-lts-bar script, is there a way to run it? Can I do `linux-lts-bar.trigger=..' or does it have to match the package name, therefore the package has to provide executable of same name as is the package name? 2022-04-10 15:52:54 mkinitfs.trigger doesn't mean run mkinitfs. it means run mkinitfs.trigger 2022-04-10 15:53:08 this program: https://gitlab.alpinelinux.org/alpine/aports/-/blob/master/main/mkinitfs/mkinitfs.trigger 2022-04-10 15:57:51 But that files does not seems to be in the package https://pkgs.alpinelinux.org/contents?branch=v3.15&name=mkinitfs&arch=x86_64&repo=main 2022-04-10 15:58:39 Ooooh, I (maybe) get it. Those are not installed anywhere and are not listed as being part of the package, but apk executes them directly from the package file or something like that, correct? 2022-04-10 16:01:08 more or less. 2022-04-10 16:01:30 the storage location of triggers, or lack thereof, is an implementation detail 2022-04-10 16:06:34 triggers are in lib/apk/db/scripts.tar 2022-04-10 16:07:30 Ah, got it, thanks everybody :) Now I just need to figure out how to compile the driver 2022-04-10 16:14:01 if you're running a script on kernel update to install a driver 2022-04-10 16:14:07 that sounds like something using akms/dkms 2022-04-10 16:14:30 which already has some integration, you can look at random *-src packages and how they define an akms script to compile out-of-tree kernel modules 2022-04-10 16:14:46 no need to hand write a shell script if it's a usual module 2022-04-10 16:22:51 how can I prevent abuild from purging all the dependencies again and again ? 2022-04-10 16:23:38 ERROR_CLEANUP in /etc/abuild.conf 2022-04-10 16:23:43 ty 2022-04-10 16:26:20 $JOBS is set automatically ? 2022-04-10 16:26:38 it's in the same file 2022-04-10 16:27:03 sigh, just built LLVM for -gnu, not -musl :| 2022-04-10 16:27:57 oh nproc does it ... 2022-04-10 16:28:57 psykose: this is in-the-tree driver, linux-lts just does not have it enabled; will lookup the akms/dkms, thanks for the tip 2022-04-10 16:29:09 what driver is it 2022-04-10 16:29:18 i am missing a command to list members of the package 2022-04-10 16:29:24 you can't use akms for in-tree i don't think 2022-04-10 16:29:29 what is a member of the package 2022-04-10 16:29:37 -DLLVM_DEFAULT_TARGET_TRIPLE="$CBUILD".. is CBUILD aarch64-linux-musl, or aarch64-alpine-linux-musl? 2022-04-10 16:30:16 echo it and see 2022-04-10 16:30:33 psykose I suppose I could untar the archive once more ... 2022-04-10 16:30:41 i have no idea what a member of the package is 2022-04-10 16:30:47 psykose: mt7921e, it is mediatek card present in some of the latest thinkpads 2022-04-10 16:30:49 do you mean the contents 2022-04-10 16:30:52 psykose: an apk 2022-04-10 16:30:56 graywolf: you could ask ncopa to add it in an usse request 2022-04-10 16:31:03 an apk of the package..? 2022-04-10 16:31:52 psykose apk installs packages/packets however you call them 2022-04-10 16:31:59 yes 2022-04-10 16:32:08 and you want the contents of an installed apk? 2022-04-10 16:32:17 psykose exactly 2022-04-10 16:32:21 apk info -L pkg 2022-04-10 16:33:24 psykose thanks at least i could find it in the manpages too :S 2022-04-10 16:33:35 it's in apk-info 2022-04-10 16:33:57 psyskose: yup .. one more thing how can I keep the distfiles ? 2022-04-10 16:34:03 ccache 1 ? 2022-04-10 16:34:56 they are already cached by default 2022-04-10 16:35:00 in /var/cache/distfiles 2022-04-10 16:35:26 ccache is a compiler cache 2022-04-10 16:35:45 and all the other apk files ? 2022-04-10 16:36:28 if you mean built packages, they are in ~/packages 2022-04-10 16:36:37 setup-apkcache 2022-04-10 16:37:38 no i mean apk's that are downloaded as binary .. /var/cache/apk doesnt hold them 2022-04-10 16:37:50 run command that psykose posted 2022-04-10 16:37:59 it will setup up cache for apks 2022-04-10 16:38:32 i see 2022-04-10 16:42:50 some problems when trying to use the stock kernel on my laptop and desktop too .. laptop showed a black screen after boot, desktop booted me into initramfs-rescue :( 2022-04-10 16:42:59 still running debian kernel ... 2022-04-10 16:45:46 guess amdgpu is still not very common .. but xf86-video-intel ? 2022-04-10 17:27:40 Hello there. 2022-04-10 17:27:49 I have a question. 2022-04-10 17:29:22 Is a root user not supposed to have permission over nobody? 2022-04-10 17:29:54 And if it does, what is the correct way of giving root such permission? 2022-04-10 17:29:59 s/user// 2022-04-10 17:29:59 Folt[m] meant to say: Is a root not supposed to have permission over nobody? 2022-04-10 17:30:05 to do what? 2022-04-10 17:30:40 To create a file in a folder owned by nobody. 2022-04-10 17:30:56 I have a mounted volume inside docker. 2022-04-10 17:31:11 is it mounted readwrite? 2022-04-10 17:31:35 root can create a file in a dir owned by nobody 2022-04-10 17:31:41 your issue is probably something else entirely 2022-04-10 17:31:51 like weird mount/bind stuff that makes what you're doing invalid 2022-04-10 17:32:21 root can't bypass a readonly mount, for instance 2022-04-10 17:32:27 not without running other stuff first 2022-04-10 17:33:17 I believe so, because when the folder was still owned by root instead of nobody, I was able to write to it. 2022-04-10 17:33:38 hmm... 2022-04-10 17:33:42 root ignores file permissions 2022-04-10 17:34:04 Okay, I'll look into the mount option then. 2022-04-10 17:37:30 do you run docker as non-root or do some other stuff that makes your containers root user not be uid 0 on the host? 2022-04-10 17:38:56 s/docker/podman/ 2022-04-10 17:38:56 Folt[m] meant to say: I have a mounted volume inside podman. 2022-04-10 17:39:01 I run podman actually 2022-04-10 17:39:13 It's run as root. 2022-04-10 17:39:34 But I'll check the mounting tutorial first before asking the question again. 2022-04-10 17:40:01 This time so I can understand it a bit better. 2022-04-10 17:45:00 just podman as root or also the containers? 2022-04-10 17:45:58 Both. I've have other issues trying to run podman as non-root and non-root inside the containers. 2022-04-10 17:46:43 This is my first attempt to get anything non-root. I use the --userns=auto option. 2022-04-10 17:48:14 I'm reading up on volumes again. The first time I tried to get it work, I was not trying to understand it. All I wanted was for the container to be created and the app to run. 2022-04-10 17:52:03 i think --userns=auto does actually remap root if you have maps in /etc/subuid 2022-04-11 10:13:21 hi 2022-04-11 10:13:36 can we get emacs 28 without native compile enabled? 2022-04-11 10:13:39 it's going to kill my laptop 2022-04-11 10:22:40 it compiles everything in /usr/share/emacs/28.1/lisp 2022-04-11 10:22:42 it's too much 2022-04-11 10:29:21 if people really need native compilation enabled, maybe setting up two different packages would be better 2022-04-11 10:29:26 i really don't like how emacs set this up 2022-04-11 10:30:03 this laptop can barely pull the source for emacs, let alone maintain keeping up with builds for it 2022-04-11 10:30:29 proper way would be to make an issue on gitlab so package maintainer can see it 2022-04-11 10:31:00 tsuki: ncopa was asking for emacs mantainer some days ago in #alpine-devel 2022-04-11 10:32:26 there's no maintainer for the emacs package? 2022-04-11 10:34:14 theorically yes but last upgrade was done by ncopa so maybe he is pretty disconnected at this momment 2022-04-11 10:34:42 i see 2022-04-11 10:35:02 i would sign up, but i don't have the hardware for it atm 2022-04-11 10:37:06 i tihnk you create an issue anyway. i currently do maintainership but i would like someone else takes over 2022-04-11 10:38:42 so make an issue under the aports project, right? 2022-04-11 10:38:50 yep 2022-04-11 10:40:43 alright. almost sun up here. couldn't sleep all night. might wait until i get some rest 2022-04-11 13:42:16 does installing bash replace /bin/sh? looking at the apkbuild Im not sure 2022-04-11 13:43:39 it doesn't 2022-04-11 13:43:41 No 2022-04-11 14:00:21 psykose: does gawk automatically get used over awk? 2022-04-11 14:00:31 yes 2022-04-11 14:00:45 you can verify with awk --version 2022-04-11 14:16:40 has anyone done benchmarking of compiling things, vs a glibc distro? 2022-04-11 14:17:02 https://groups.google.com/g/git-for-windows/c/9WrSosaa4A8 2022-04-11 17:18:29 Hello all 2022-04-11 17:18:38 So,this is official Group,right? 2022-04-11 17:18:47 yes 2022-04-11 17:18:59 I have come from Telegram unofficial group, greetings 2022-04-11 17:19:08 welcome! 2022-04-11 17:21:09 I have been thinking of using alpine as daily driver OS. minimal stable nature of alpine was attractive to me. thanks to bubblewrap, it lets me run any glibc dependant program on alpine 2022-04-11 17:22:18 my one question is about init. i don't belong to any cult but i wanted to know progress of s6 on alpine and frontend of s6 that was in development 2022-04-11 17:24:34 skarnet is working on it, but haven't heard any progress yet 2022-04-11 17:25:01 S6 will be integrated in Alpine once s6-rc and s6-frontend are done 2022-04-11 17:25:38 so whats so good about s6 that mandates a change? 2022-04-11 17:25:51 except that it isnt systemd =) 2022-04-11 17:25:51 i was looking at how to at least get started with some *-s6 subpackage as an example of how to even write a script, but i could not figure out what the policy on anything would be, and there is no frontend yet anyway 2022-04-11 17:26:02 it actually supervises things 2022-04-11 17:26:05 it's more, "what is so wrong about openrc that mandates a change" 2022-04-11 17:26:16 panekj: and the answer is... ? 2022-04-11 17:26:22 so,openrc will be phased off? 2022-04-11 17:26:28 its like openrc but it will monitor the services? 2022-04-11 17:26:36 for me, it's missing user services 2022-04-11 17:26:51 how do you mean? 2022-04-11 17:26:53 i should set up a user s6 tree sometime.. 2022-04-11 17:26:58 me too 2022-04-11 17:27:14 Apachez: systemd --user equivalent 2022-04-11 17:27:57 Ermine: seems skarnet hasn't worked on s6-frontend this year 2022-04-11 17:28:05 it would be nice for my supervisor to manage my notification daemon, audio daemon, etc. etc. 2022-04-11 17:28:50 s6 is process supervision suite. It launches processes and restarts them if they die. Also it manages logs 2022-04-11 17:29:44 s6-rc is service manager, which runs on top of s6 and manages system state 2022-04-11 17:30:14 so more attacksurfaces then 2022-04-11 17:32:04 varbhat: he will eventually 2022-04-11 17:33:11 Apachez: no, situation is opposite: responsibilities are split between different packages, which means less code in each of them, which means less attack surface 2022-04-11 17:33:21 psykose: me too 2022-04-11 17:38:07 s6-rc is 'equivalent' to rc-service 2022-04-11 17:38:21 you manage services with it 2022-04-11 17:38:24 what attack surface 2022-04-11 17:38:48 you're right, we should have init only and define every service in /etc/inittab 2022-04-11 17:38:51 attack surface solved 2022-04-11 17:44:28 psykose: or sinit 2022-04-11 19:22:13 i'm looking for a way to reload libvirtd 2022-04-11 19:22:22 but it's not supported as an arg for /etc/init.d 2022-04-11 19:22:26 recommendations? 2022-04-11 19:26:28 You can try to send SIGHUP to it and see whether it worked 2022-04-11 19:28:40 i'll try that 2022-04-11 19:33:04 that would mean someone would have to add it to the init.d script, if it works 2022-04-11 19:45:44 i'll report back 2022-04-11 20:27:31 opviel: did it work for you? 2022-04-11 20:28:49 Ermine: nah... doesn't do anything to the process 2022-04-11 20:29:20 according to the manual sighup makes it reload config 2022-04-11 20:29:42 so either something is broken or you sent it to the wrong pid 2022-04-11 20:29:50 and libvirtd unit file does the same thing 2022-04-11 20:30:39 you can try start-stop-daemon --signal HUP --pidfile /run/libvirtd.pid 2022-04-11 20:54:56 alright 2022-04-11 20:55:02 i 2022-04-11 20:55:06 'll try further 2022-04-11 20:55:07 thank you 2022-04-11 22:35:59 hi, noob question: how do you set up wifi in the installer? wlan0 interface doesn't show up after i install firmware and run modprobe. 2022-04-11 22:37:35 you need to use the extended image to have wpa_supplicant installed i think 2022-04-11 22:37:45 otherwise uhh.. you can't 2022-04-11 22:38:14 i am using the extended image. but i think wpa_supplicant requires the interface to exist in the first place, doesn't it? 2022-04-11 22:38:23 ah, that's what you mean 2022-04-11 22:38:37 what's in dmesg? the module loaded should make the interface show up 2022-04-11 22:40:38 on boot, "Direct firmware load for iwlwifi-blahblah.ucode failed with error -2". there is no new message in dmesg after modprobe 2022-04-11 22:45:03 that means you're missing the firmware 2022-04-11 22:45:12 the.. linux-firmware in the .iso is not complete 2022-04-11 22:46:04 i think all those iwlwifis are in firmware-other 2022-04-11 22:46:18 which is.. 100mb as a compressed apk, and not included 2022-04-11 22:46:31 you wouldn't be the first person to run into this 2022-04-11 22:46:52 i don't remember what is or isn't included in either .iso from the full set of firmware 2022-04-11 22:46:56 i installed firmware-other and it does contain the exact file that dmesg was complaining about 2022-04-11 22:47:11 yep 2022-04-11 22:47:23 now with it present reloading the module should work 2022-04-11 22:47:53 nope 2022-04-11 22:48:11 mm 2022-04-11 22:48:35 are you sure it's the correct module and that you rmmod'd it first 2022-04-11 22:48:44 i don't remember how to fully make it reload the firmware actually.. 2022-04-11 22:49:06 i think it might be because i remounted /.modloop read-write following these steps: https://wiki.alpinelinux.org/wiki/Configure_a_Wireguard_interface_(wg)#Running_with_modloop 2022-04-11 22:49:33 without it, apk fails to install firmware-other since it's read-only 2022-04-11 22:49:39 that should probably still be fine if the file locations are the same 2022-04-11 22:49:45 it just needs to read a firmware file 2022-04-11 22:49:48 and fails otherwise 2022-04-11 22:50:56 i didn't run rmmod before. now that i run it, i get a Resource temporarily unavailable error 2022-04-11 22:51:56 Try to reboot, install firmware again and modprobe 2022-04-11 22:55:45 it worked! thank you very much 2022-04-11 22:55:55 rmmod was the problem 2022-04-12 03:15:29 Anyone hosting xRDP with X (XFCE)? I believe I'm in all the proper groups and have all the requisite services running, but I can't get any hosts to connect to xRDP session manager. Just a black screen with a pointer, no interaction. 2022-04-12 03:15:29 This a side-effect of rootless XORG patches landing? 2022-04-12 15:08:54 Hi everyone! A question regarding kernel updates in diskless mode. 2022-04-12 15:09:36 The docs say that since apk upgrade doesn't upgrade the kernel, one must do a full upgrade. 2022-04-12 15:09:51 where does it says that? 2022-04-12 15:10:02 https://wiki.alpinelinux.org/wiki/Raspberry_Pi#Update_the_System 2022-04-12 15:10:03 here 2022-04-12 15:10:19 under "Post Installation", Update the System, last line is a note 2022-04-12 15:10:59 I was wondering, what about if the kernel gets a patch before the next minor release. What's the best way to update it. 2022-04-12 15:11:21 Extract the file from the apk and manually copy it to the boot media? 2022-04-12 15:12:24 You can upgrade just the kernel package 2022-04-12 15:15:20 hm, well, for start, it's not installed by default; so you'd suggest to install it, and then it gets normally upgraded, so I just copy the kernel image to the boot media? 2022-04-12 15:38:57 okay, not so simple. The default deployment uses the modloop, whereas the kernel package provides a MZ kernel image. So a bit more tinkering would be required than simply copying the files (or symlinking the /boot to /media/...) 2022-04-12 15:40:41 if it wasn't obvious from start, I'm doing this on a raspi :) 2022-04-12 16:07:51 There is a tool called update-kernel 2022-04-12 16:10:51 thanks, I was given a hint about it via /msg ;) checking now - it seems I was looking at the wrong part of wiki and this is all explained in Package management 2022-04-12 16:11:34 I'm verifying if the 8 GB requirement can be circumvented by the TMPDIR var 2022-04-12 16:51:40 in theory 2022-04-12 17:16:38 doas seems to be broken on edge, im getting error on root and user that's added to doas.conf 2022-04-12 17:17:53 post the doas.conf and the shell output 2022-04-12 17:17:57 doas: Operation not permitted 2022-04-12 17:18:14 doesn't show what you typed 2022-04-12 17:18:20 doas uname 2022-04-12 18:02:41 can you reproduce this? http://ix.io/3UYU 2022-04-12 18:03:15 run it as a script, not directly on the command line 2022-04-12 18:03:27 uncomment each of the wget lines in turn 2022-04-12 18:07:57 seems to be an alpine bug rather than a busybox bug, because I get the same if I use dash and nawk 2022-04-12 18:10:40 I do not get the broken behavior on void with glibc 2022-04-12 18:11:29 broken one is this if the link dies: wget -qSO f "https://alpinelinux.org" 2>&1 | awk "{ exit }" 2022-04-12 18:13:11 hmm actually you don't have to run it from a script, I'm getting the same size 0 file when run interactively too 2022-04-12 18:13:54 maybe it has relation with busybox/ssl_client 2022-04-12 18:14:19 this is also broken: wget -qSO f "https://alpinelinux.org" 2>&1 | awk "{ print; exit }" 2022-04-12 18:14:28 donoban: you can reproduce? 2022-04-12 18:14:59 they all work for me 2022-04-12 18:14:59 well, I run behind a proxy which will complicate the things more 2022-04-12 18:15:34 shouldn't be a busybox issue because I get the same when running this from dash: wget -qSO f "https://alpinelinux.org" 2>&1 | nawk "{ print; exit }" 2022-04-12 18:15:36 I get HTTP/1.1 301 Moved Permanently 2022-04-12 18:15:52 idk what the awk is supposed to be for, but it would have nothing to do with the final output file size 2022-04-12 18:16:00 it would be 0 if the request failed 2022-04-12 18:16:02 donoban: and what size is f? 2022-04-12 18:16:20 there is no f 2022-04-12 18:16:28 oh 2022-04-12 18:16:33 well that's weird on its own lol 2022-04-12 18:16:38 might be the proxy thing 2022-04-12 18:16:53 if its giving you 301 2022-04-12 18:16:55 I'm getting 200 2022-04-12 18:17:16 but the file size is still 0, unless I remove the "exit" from the awk command or the -S from the wget command 2022-04-12 18:17:22 I remember that busybox wget over proxy was broken for https 2022-04-12 18:17:40 finally gnu wget was added as a workaround for setup-alpine over a proxy 2022-04-12 18:17:57 ah, this is GNU wget 2022-04-12 18:18:04 anyway, what do you want to achieve? 2022-04-12 18:18:11 I'm debugging a script 2022-04-12 18:18:35 the script fetches pages, and today it started giving me files of size 0 2022-04-12 18:18:53 so I got down to this minimal example which I don't understand 2022-04-12 18:19:07 i can't reproduce it at all 2022-04-12 18:19:32 psykose: what size is f? 2022-04-12 18:19:39 the whole webpage 2022-04-12 18:19:48 so like how big 2022-04-12 18:19:50 8.3k 2022-04-12 18:20:33 uhM, it seems that I get "too many redirections" 2022-04-12 18:20:35 yea that's what I get with the ones labeled "WORKS" 2022-04-12 18:20:43 i get that with every single one 2022-04-12 18:20:49 huh 2022-04-12 18:21:13 http://ix.io/3UYZ 2022-04-12 18:21:59 should be a porlbem with the proxy? 2022-04-12 18:22:22 yea must be, I'm not getting that at all 2022-04-12 18:22:30 same results with gawk or bb awk, gwget or bb wget 2022-04-12 18:22:51 well I could reboot but maybe it would be more valuable to keep this bug alive in order to investigate 2022-04-12 18:24:24 have you tried with curl? 2022-04-12 18:26:36 uname -a shows Linux styx 5.15.16-0-virt #1-Alpine SMP Wed, 26 Jan 2022 16:46:12 +0000 x86_64 GNU/Linux 2022-04-12 18:26:49 psykose: what's yours? 2022-04-12 18:27:02 Linux sumire 5.17.1-0-xyz-rt #1-xyz-rt SMP PREEMPT Sun, 10 Apr 2022 13:58:36 +0000 x86_64 Linux 2022-04-12 18:28:04 it's most probably not a kernel issue 2022-04-12 18:34:41 donoban: I couldn't translate the wget options to curl, but various curl commands all work as expected 2022-04-12 18:35:31 psykose: maybe a libc issue 2022-04-12 18:35:36 I just tried curl and it gets the page properly 2022-04-12 18:35:41 does not seem like a libc issue either 2022-04-12 18:35:43 yeah 2022-04-12 18:35:50 I don't know why wget has this redireciton problem 2022-04-12 18:36:12 yeah that's weird 2022-04-12 18:36:24 I should check the headers 2022-04-12 18:40:19 here is a transcript so you all believe me http://ix.io/3UZ5 2022-04-12 18:40:26 no idea what to try though 2022-04-12 18:43:06 could it be ipv6 related? 2022-04-12 18:43:51 scratch that 2022-04-12 18:44:08 Apachez: can u reproduce? 2022-04-12 19:06:11 How do I boot from a serial console on the first boot? The wiki suggests editing /boot/extlinux.cfg, but I've no idea how to do that. I'm using the ISO image available on the website. 2022-04-12 19:06:29 E.g.: I've no idea how to edit a file in the ISO image before having booted. 2022-04-12 19:33:16 You'd need to remake the iso https://www.unixmen.com/edit-iso-files-using-mkisofs-in-linux/ 2022-04-12 19:35:59 You can add it as a kernel parameter 2022-04-12 19:38:18 ikke: I'm guessing they would need a normal console to edit the cmdline in the Syslinux menu during boot...or does Syslinux automatically handle serial as well as normal console? 2022-04-12 19:39:02 Yes, it should 2022-04-12 19:39:48 That's how I did it, boot the iso, changed the kernel parameters to set the console to a serial device, and then I could install alpine 2022-04-12 19:46:47 But I don't want to install alpine, I want to run the ISO in a VM. 2022-04-12 19:47:02 I want to install some tools on that image, generate some builds, and then scrap it. 2022-04-12 19:47:57 WhyNotHugo: then why would you be using the serial in the vm? 2022-04-12 19:48:12 ran apk -U upgrade and I'm still having the issue with wget -qSO f "https://alpinelinux.org" 2>&1 | awk "{ print; exit }" 2022-04-12 19:48:13 I can copy-paste into the console very easily. 2022-04-12 19:48:24 wget is GNU Wget 1.21.2 built on linux-musl. 2022-04-12 19:49:00 in a vm you don't need a "serial" console.. the vm's console is already accesible 2022-04-12 19:49:07 bleb: Why not curl? 2022-04-12 19:49:12 you can copy/paste there 2022-04-12 19:49:15 Tenkawa: through what medium, if not serial? 2022-04-12 19:49:25 Tenkawa: How? 2022-04-12 19:49:43 WhyNotHugo: I couldn't figure out how to replicate wget -qSO with curl 2022-04-12 19:50:02 bleb: every vm software I know of has console redirection 2022-04-12 19:50:05 and wget should work anyway 2022-04-12 19:50:18 I'm using virt-manager. What console if not serial 2022-04-12 19:50:19 ? 2022-04-12 19:50:26 Vmware/Parallels/UTM 2022-04-12 19:50:33 Qemu 2022-04-12 19:50:44 they all can redirect mouse output 2022-04-12 19:50:53 Tenkawa: yeah... but serial won't work if its not enabled in the OS 2022-04-12 19:50:57 I don't understand what other console you mean; I've only seen the tty console. 2022-04-12 19:51:14 bleb: Right! So what should I be using if I want that enabled? 2022-04-12 19:51:27 Do I actually want the extended image? https://alpinelinux.org/downloads/ 2022-04-12 19:52:03 sounds like whoever built the ISOs thought it didn't matter if serial was disabled 2022-04-12 19:52:12 so u have to build the ISO yourself 2022-04-12 19:52:17 or alpine has to do better 2022-04-12 19:54:13 Okay, I'll build my own ISO and see if the change can be upstream unobstrusively. 2022-04-12 19:55:13 sometimes the bootloader lets you add kernel parameters via serial but that may not be the case for alpine 2022-04-12 19:55:22 and even so it's a pain to have to do that every time you boot the ISO 2022-04-12 19:59:24 I think I'll script the ISO generation and try it in sourcehut's builders. I need a "first" alpine runtime to bootstrap into the OS. 2022-04-12 20:02:37 You can make a writable boot medium and adjust the configuration 2022-04-12 20:03:57 here's another command line which barely works: 2022-04-12 20:03:58 wget -qSO f "https://alpinelinux.org" 2>&1 | awk "{ print } NR > 10 { exit }" && ls -l f 2022-04-12 20:04:08 if you change the 10 to 9, it stops working 2022-04-12 20:04:46 so it seems like a wget bug when the pipe doesn't consume all of the output from -S 2022-04-12 20:09:48 psykose: which version of wget did u have? 2022-04-12 20:10:18 a more distilled example: wget -O f "https://alpinelinux.org" 2>&1 | true 2022-04-12 20:11:03 bleb: what happens for you when you execute that? 2022-04-12 20:12:24 ikke: f becomes an empty file 2022-04-12 20:12:39 without | true, f contains the page 2022-04-12 20:13:57 if you remove 2&>1 it works as well 2022-04-12 20:13:58 strange 2022-04-12 20:14:03 that one is reproducible 2022-04-12 20:14:15 ah good 2022-04-12 20:14:31 with '| true' it does not create 'f' for me at all 2022-04-12 20:16:44 actually I can reproduce the latest one on glibc void as well 2022-04-12 20:16:46 without 2&>1 it does 2022-04-12 20:16:54 so this could be two different bugs... 2022-04-12 20:18:47 with the latest example it makes a bit more sense for it to mess up, because it's trying to write the progress bar 2022-04-12 20:20:21 this example is more concerning and can still be reproduced outside my alpine vps: wget -qSO f "https://alpinelinux.org" 2>&1 | true 2022-04-12 20:22:37 i'm not sure this qualifies as a bug 2022-04-12 20:23:26 yeah theres no real way you'd run into it 2022-04-12 20:23:41 this one on the other hand is what I actually ran into: wget -qSO f "https://alpinelinux.org" 2>&1 | awk '{ print; exit }' 2022-04-12 20:23:58 but I can't reproduce that on my void, only on alpine 2022-04-12 20:24:20 if you close stdout, wget will receive SIGPIPE. in order to not close, it would have to specifically ignore it 2022-04-12 20:25:07 if wget doesn't ignore SIGPIPE, then it comes down to a race condition 2022-04-12 20:25:26 whats the race condition 2022-04-12 20:25:53 hm, actually, SIGPIPE is sent synchronously 2022-04-12 20:25:59 oh, it's probably related to stdio buffering 2022-04-12 20:26:44 by void you mean glibc void? 2022-04-12 20:26:47 yes 2022-04-12 20:27:01 awk is also line buffered though 2022-04-12 20:27:10 so its not like its sending SIGPIPE in the middle of a line 2022-04-12 20:27:22 wget output is (probably) not line-buffered 2022-04-12 20:27:49 I guess it could receive SIGPIPE after partly writing a line to the pipe which hasn't been consumed by awk 2022-04-12 20:28:09 maybe strace? 2022-04-12 20:30:07 if I prepend strace I only get this: 2022-04-12 20:30:08 execve("/usr/bin/wget", ["wget", "-O", "f", "https://alpinelinux.org"], 0x7fff639aa1b8 /* 23 vars */) = 0 2022-04-12 20:30:32 add -f 2022-04-12 20:31:06 oops I mean this 2022-04-12 20:31:09 execve("/usr/bin/wget", ["wget", "-qSO", "f", "https://alpinelinux.org"], 0x7ffdabb6d208 /* 23 vars */) = 0 2022-04-12 20:31:25 strace -f gives the same thing ^ 2022-04-12 20:33:45 the strace output is going into the pipe 2022-04-12 20:34:07 add -o foo.strace 2022-04-12 20:41:28 heres the result of strace -o strace.out wget -qSO f "https://alpinelinux.org" 2>&1 | awk '{ print; exit }' 2022-04-12 20:41:40 http://ix.io/3UZD 2022-04-12 20:41:47 (^ the strace.out) 2022-04-12 20:43:44 lol, if I run it on glibc void, the strace causes the bug to be reproduced 2022-04-12 20:43:47 writev(2, [{iov_base=" Server: nginx\n", iov_len=16}, {iov_base=NULL, iov_len=0}], 2) = -1 EPIPE (Broken pipe) 2022-04-12 20:43:53 without the strace I can't reproduce the bug 2022-04-12 20:44:54 maybe some of you can reproduce it using strace 2022-04-12 20:46:39 I wanted to compare the strace to a working run on glibc void, but strace makes it fail the same as on alpine 2022-04-12 21:41:41 meaning it's almost certainly a race condition 2022-04-12 22:06:21 When apk executes man-db trigger, I get following messages: mandb: iconv_open ("UTF-8//IGNORE", "ISO-8859-1"): Invalid argument 2022-04-12 22:07:47 sounds like a Latin 1 man page to me 2022-04-12 22:08:21 iirc musl didnt do //IGNORE? 2022-04-12 22:10:01 Those pages are /usr/share/man/man1/grolbp.1.gz and /usr/share/man/man1/mmroff.1.gz 2022-04-12 22:16:35 Well, do they contain non-ASCII Latin 1 characteres? 2022-04-12 22:18:05 Yes 2022-04-12 22:18:18 zless shows such 2022-04-12 22:18:46 If so, that's a bug in those. 2022-04-12 22:22:11 So, I should report to groff that these man pages shouldn't contain non-ASCII characters? 2022-04-12 22:24:32 Check the groff sources first. If they are wrong, report to groff@. They should contain either ASCII or perhaps UTF-8. 2022-04-12 22:27:30 Ok, thank you 2022-04-12 22:28:54 Hah, in HEAD it's "fixed" by using even less portable constructs. 2022-04-13 02:36:50 hello 2022-04-13 02:38:00 i am running alpine linux on my raspberry pi 2b, and ran into the "no space left on device" error while trying to browse the web. 2022-04-13 02:38:24 i rebooted the system, but it could only put me in the recovery shell 2022-04-13 02:38:55 also my persistent storage volume seems to be gone along with my packages, and it seems like there is no package database 2022-04-13 02:39:14 recovery shell usually does not have the disk mounted 2022-04-13 02:39:28 hm 2022-04-13 02:43:31 i can access one partition of my sd card, but not any others 2022-04-13 02:44:07 i'll look at the documentation for /etc/fstab 2022-04-13 02:46:48 according to fdisk, only one partition is recognizable, and i can access it from /media/mmcblk0p1 2022-04-13 02:48:45 there should be /media/mmcblk0p1/persist.img and a /media/persist, but neither exist 2022-04-13 02:50:20 sounds like your storage died 2022-04-13 02:50:27 i found some ovl files in /tmp 2022-04-13 02:51:04 which has system configs for some packages i installed 2022-04-13 02:51:24 and apparently there's also an apk ovl file in the sd card partition 2022-04-13 02:51:47 > sounds like your storage died 2022-04-13 02:51:47 the persistent storage? 2022-04-13 02:52:59 dunno 2022-04-13 02:53:17 either the thing can't read the partition for some reason or it's completely broken in some sense 2022-04-13 03:11:32 so i've decided to look at the ovl archive 2022-04-13 03:11:42 but it's been sitting on it for way too long 2022-04-13 03:11:54 i just ran `tar t archive.tar.gz` 2022-04-13 14:15:06 hi, somehow I can't get input in X.org on qemu-system-x86_64 -M q35 -vga vmware, using linux-virt kernel, xf86-input-{evdev, libinput, vmmouse} and xfce4 are installed 2022-04-13 14:15:19 and udev 2022-04-13 14:15:49 oscareczek: is the user part of the input group? 2022-04-13 14:16:21 now it is, didn't help 2022-04-13 14:16:57 did you login again? 2022-04-13 14:17:03 oh, reboot helped 2022-04-13 14:17:29 I knew it was something obvious, that's about this issue 2022-04-13 14:17:54 however, I use -vga vmware because -vga virtio doesn't detect screens and -vga qxl segfaults on launch 2022-04-13 14:19:17 I'd certainly prefer the first one since you can freely resize QEMU window and screen resolution will follow, but wiki prefers QXL for some reason 2022-04-13 14:21:13 I'm on edge BTW 2022-04-13 16:27:33 could someone confirm that installing linux-firmware-* apks on diskless also requires update-kernel? 2022-04-13 16:55:18 waydot: last time I checked, installing linux-firmware with modloop didn't work at all 2022-04-13 16:55:49 but my *diskless* is PXE-booted, so it may be different than on a RAM-based install on an rpi or smth 2022-04-13 17:02:30 is anyone aware of how to get the linux kernel debug symbols for alpine? 2022-04-13 17:05:28 they don't exist i don't think 2022-04-13 17:05:38 unless you are talking about the System.map that doesn't really work 2022-04-13 17:06:10 sdomi: thanks, yeah, I've got rpi, but I think I might be on the wrong path here, still investigating 2022-04-13 17:11:25 psykose: is there a way to generate them or is that basically "recompile the kernel" 2022-04-13 17:12:02 and/or: is there a good way to request for such a package to be added to the main repositories? 2022-04-13 17:12:52 for linux-kernel specifically i don't have a single clue how a debug package would work 2022-04-13 17:12:59 create an issue on gitlab and ask ncopa about it 2022-04-13 17:14:50 yeah, I believe normally this is done as part of the build process for the regular kernel package, so you basically build the regular kernel package and the debug package at the same time, copying just the debug symbols from the kernel build results into the debug symbols package. ill create an issue. 2022-04-13 17:24:06 hi, does anybody know to get logs from networkmanager? 2022-04-13 17:25:04 IIRC they're under /var/log/NetworkManager..? I may be misremembering tho, can't check rn 2022-04-13 17:25:10 doesn't it just log to syslog? 2022-04-13 17:25:33 it's in var/log/messages 2022-04-13 17:25:45 by default 2022-04-13 17:25:48 you can also kill the service and launch NetworkManager with the -n switch, it prints them to stderr then 2022-04-13 17:26:44 if you want to see what's an interface doing, you can run `nmcli d monitor ethX` 2022-04-13 17:27:00 thanks 2022-04-13 17:27:44 or without ethX, it will monitor all interfaces 2022-04-14 00:07:59 hi, wifi doesn't connect for some reason upon install. tried standard and extended isos. 2022-04-14 00:09:23 CPU: 8-Core AMD Ryzen 7 4800H with Radeon Graphics (-MT MCP-) speed/min/max: 1397/1400/2900 MHz 2022-04-14 00:09:23 Kernel: 5.13.0-39-generic x86_64 Up: 8m Mem: 1590.3/15416.7 MiB (10.3%) Storage: 505.76 GiB (58.4% used) Procs: 333 2022-04-14 00:09:23 Shell: Zsh 5.8 inxi: 3.3.06 2022-04-14 00:38:02 could you post dmesg 2022-04-14 00:39:01 crystal ball says broadcom 2022-04-14 00:40:00 bigger crystal ball says missing firmware load 2022-04-14 00:43:15 also i'm not sure what iso comes with a 5.13 kernel 2022-04-14 00:43:41 5.13.0-39-generic doesn't sound like alpine 2022-04-14 00:43:42 thats what im running now, not the alpine 2022-04-14 00:45:30 do you still want my dmesg for my current distro? 2022-04-14 00:47:03 dmesg of the one that fails 2022-04-14 00:50:16 ok well ill have to go back and try it again then type it or save it somehow. isnt it long? and i couldnt connect to net at all 2022-04-14 00:50:20 no wth port on computer. 2022-04-14 00:50:22 eth 2022-04-14 00:51:35 dhcp couldnt discover ip, so did ifconfig eth0 up, then udhcpc eth0 from a guide, still nothing 2022-04-14 00:52:19 if there is no ethernet port on the laptop then what is eth0 2022-04-14 00:52:51 idk 2022-04-14 00:53:00 what should i have done 2022-04-14 00:53:51 ah there is one, no eth ports where i am, though 2022-04-14 00:58:24 how would udhcpc eth0 get you a dhcp lease/networking if the port is not even connected..? 2022-04-14 00:58:35 regardless, i can't tell you why the wifi doesn't work without seeing the dmesg 2022-04-14 00:58:52 it's probably missing firmware, or something else related to the specific card 2022-04-14 00:59:04 idk, isnt there wifi connection like other distros. i dont know these things 2022-04-14 00:59:05 and even with it there, you would have to use wpa_supplicant manually 2022-04-14 00:59:15 would outting eth cord in router make it work 2022-04-14 00:59:23 probably 2022-04-14 05:03:41 FREENODE AND LIBERA MOVE TO irc.worldhacker.org is open for public #0day , [ 0day (xc) Our ] WhiteHat Hacker Team And Dev Official Public Channel : https://worldhacker.org , REQUEST PROGRAM , SOURCE CODE , OR ANYTHING HERE ... . NEED STORAGE OR VPS ??? Free 2022-04-14 05:58:28 sure buddy 2022-04-14 12:20:26 I came, I saw, I lol'd 2022-04-14 15:02:11 Trying to move /var to a new partition. Started alpine in single user mode. Can't figure out how to mount the filesystem. fdisk -l shows nothing. 2022-04-14 15:20:20 charmango: mount with -o remount,rw perhaps? 2022-04-14 15:27:11 huh, mkinitfs does single before mounting root. i don't think this is how historical single user mode worked 2022-04-14 15:27:43 charmango: in general i recommend using separate bootable medie instead of single-user mode 2022-04-14 15:28:57 ok 2022-04-14 15:29:58 if you insist on using single then i think you can run nlplug-findfs -p /sbin/mdev $KOPT_root like in https://gitlab.alpinelinux.org/alpine/mkinitfs/-/blob/master/initramfs-init.in#L513 2022-04-14 15:30:24 nlplug-findfs invokes mdev which loads modules 2022-04-14 15:34:27 Hello71, that worked. Thank you. 2022-04-14 15:35:04 np 2022-04-14 16:48:49 Anyone on Edge using libvirt/qemu? Many of the services are failing/crashing due to "No permission to apply cgroup settings" 2022-04-14 16:49:01 This is new as this was working prior during the last week or so 2022-04-14 17:00:28 ACTION sent a code block: https://matrix.org/_matrix/media/r0/download/matrix.org/qPenOEZkzEDbyMPDaIUBCrEc 2022-04-14 17:03:05 Should I just disable CGROUP support entirely in /etc/rc.conf? 2022-04-14 17:03:12 I have it set to hybrid 2022-04-14 18:23:23 hi, alpine doesn't detect my partitions? only whole drive, nvme0n1 2022-04-14 18:24:01 Daniel123: Did you set them up with Alpine's setup scripts, or are you using existing partitions alongside Alpine? 2022-04-14 18:24:15 existing 2022-04-14 18:24:29 How are they setup? LUKS, LVM, anything else like that? 2022-04-14 18:24:47 And what FS? 2022-04-14 18:24:58 no, ext4, home on seperate 2022-04-14 18:28:06 Hmm... Not sure then. Are you using a DE, or are you terminal? Do you have things like udisks2? 2022-04-14 18:35:57 on alpine or my distro 2022-04-14 18:36:09 i used gparted 2022-04-14 18:38:00 what does "on alpine or my distro" mean 2022-04-14 18:38:14 what did your question mean 2022-04-14 18:56:38 ?? 2022-04-14 18:56:54 alpine instal is in vte so no i dont have a dektop on that 2022-04-14 19:00:44 I was asking if you had a DE in Alpine since they pull in a bunch of other packages which might make mounting the partitions easier 2022-04-14 19:00:59 udisks2 being one of the main ones I can think of highlighted by the Wiki on setting up a DE in Alpine 2022-04-14 19:02:40 oh , but don ti have to install alpine first to set up a de 2022-04-14 19:04:29 Ah, I was not understanding. So you're trying to install Alpine alongside an existing install on disk? 2022-04-14 19:16:40 Daniel123: What does "doesn't detect my partitions" mean? 2022-04-14 19:17:09 Do they not appear in /dev? Do you use mdev? udev? 2022-04-14 19:19:41 #trascio 2022-04-14 19:21:04 no thanks 2022-04-14 19:22:00 unlucky 2022-04-14 19:26:26 Saijin_Naib: ys 2022-04-14 19:26:28 yes 2022-04-14 22:51:58 Would it be a complete disaster if I just chown'd the whole cgroup directory tree? I need this working... 2022-04-14 22:55:45 Daniel123: And you're positive there is nothing funky going on like LVM, LUKS, MDAM, etc? If so, I can't think of a reason why Alpine wouldn't be able to see anything short of maybe a missing kernel module to access your NVME drive, but if I recall, that should have been sorted in the linux-lts kernel a long time ago. What version of the installation media are you using? Current 3.15.3? 2022-04-14 23:33:06 3.15.4. also tried with 3.14 2022-04-14 23:33:16 can i find the missing module somewhere 2022-04-14 23:35:41 still have windows bootloader even though i deleted the os from the computer, could that be an issue? 2022-04-14 23:53:04 That shouldn't stop Alpine from seeing the partitions, though that might not be ideal later on 2022-04-14 23:53:10 Not sure what module might be missing 2022-04-14 23:59:38 https://github.com/containers/crun/issues/696#issuecomment-870343322 2022-04-14 23:59:54 this but substituting /sys/fs/cgroup/openrc instead of systemd seems to be the ticket 2022-04-15 00:00:06 Should this not be something setup automatically by OpenRC/Alpine at install? 2022-04-15 00:07:19 Nope, re-owned by root after reboot. 2022-04-15 00:07:58 i don't understand why it needs to be user owned 2022-04-15 00:09:43 at first it wasn't there, then when it is it works with root 2022-04-15 00:09:46 rootless is a separate issue 2022-04-15 00:10:23 https://github.com/containers/crun/pull/698/files then they added this 2022-04-15 00:10:34 so.. yeah, seems to work as intended 2022-04-15 00:10:55 not sure if it still applies given the age, but it was only a rootless issue and they only supported legacy v1 cgroups 2022-04-15 00:13:17 seems to support v2 now as well https://github.com/containers/crun/blob/main/crun.1.md#cgroup-v2= 2022-04-15 00:13:48 since 0.15 2022-04-15 00:13:53 hm, weird 2022-04-15 00:14:11 i guess it was the cgroup configuration then 2022-04-15 00:17:14 Hmm, thanks for the pointers 2022-04-15 00:17:36 I added my username to the /etc/subuid and /etc/subgid files as per podman's rootless config, but no love yet 2022-04-15 00:17:57 I'm mystified as to how I managed to screw this up overnight, but here I am haha 2022-04-15 00:30:14 rootless is hard to set up 2022-04-15 00:30:22 i don't think i ever managed 2022-04-15 00:30:29 but i also never really use containers, just for some light testing 2022-04-15 00:30:37 so regular doas .. run is fine to me 2022-04-15 00:31:17 Yeah, I wasn't expecting to need to delve into any of this, but out of the blue, libvirt/virt-manager can no longer start, throwing the cgroup permission error 2022-04-15 00:31:35 Hence trying to setup rootless to get those services to no longer crash at boot 2022-04-15 00:42:40 I just need virt-manager running so I can use my Windows 11 image for ArcGIS/Office 2022-04-15 01:10:53 well libvirtd runs as root 2022-04-15 01:11:17 then iirc for virt-manager to work.. i needed to copy in some dbus definition since our packaging was broken (otherwise you need root) 2022-04-15 01:11:33 this seems unrelated to crun and containers 2022-04-15 01:11:57 or at least none of those things to run windows even pull in crun 2022-04-15 01:13:42 Hmm, I wonder what changed in this week's edge updates to break a working kvm/qemu/libvirt install I've been using for over a week 2022-04-15 01:14:07 I wish I knew what to debug so I could actually make progress on fixing this instead of just random tweaks that seem right 2022-04-15 01:14:37 If I'm already part of the kvm, libvirt, qemu groups do I still need root or that dbus hack? 2022-04-15 01:53:20 the dbus/polkit/something/iforgot thing didn't check for group membership 2022-04-15 01:53:31 as in, it was missing a check for it to let you do stuff 2022-04-15 01:53:47 Ahhh, gotcha 2022-04-15 01:53:56 So, unresolved issue still likely? 2022-04-15 01:54:01 I'll poke and see what I can find 2022-04-15 01:54:44 i'll try and look in a se 2022-04-15 01:54:45 sec 2022-04-15 01:55:22 Nah, you're slammed with builder stuff 2022-04-15 01:55:28 You've already helped a lot πŸ™‚ 2022-04-15 01:57:40 https://img.ayaya.dev/gphL3SQTYYwj.png simple start being in all the groups, after adding qemu-system-x86_64 libvirt-daemon virt-manager 2022-04-15 01:58:02 then adding polkit and starting it.. and it works 2022-04-15 01:58:21 except, when adding a new machine, there is no 'active connection' 2022-04-15 01:59:43 then, starting libvirtd, and adding the default qemu:///system connection works 2022-04-15 02:00:33 i had to restart libvirtd after polkit, but that's it 2022-04-15 02:00:41 works as regular user and i can create a machine it seems 2022-04-15 02:01:45 mm 2022-04-15 02:01:50 there is virt manager 4 now 2022-04-15 02:01:52 hehe 2022-04-15 02:06:21 https://img.ayaya.dev/9fjBmdXqkvbC.png 2022-04-15 02:06:22 and poof 2022-04-15 02:06:34 this is with 4 even which i just built to test, but it works on 3.2 as well 2022-04-15 02:07:08 didn't do anything but add to groups, su myself, add virt-manager libvirt-daemon qemu-system-x86_64 polkit qemu-hw-usb-redirect qemu-ui-spice-app qemu-audio-spice qemu-ui-spice-core qemu-chardev-spice, start polkit/libvirtd 2022-04-15 02:10:10 Awesome, thank you! 2022-04-15 02:10:26 So it looks like I need to tune OpenRC to make polkit initialize later than the others... 2022-04-15 02:10:42 Guess I can bump the *virt/*qemu stuff into the boot runlevel instead of default so they initialize before polkit 2022-04-15 02:13:58 not at all 2022-04-15 02:14:04 before, if anything 2022-04-15 02:14:07 but i don't think it will matter 2022-04-15 02:14:35 Hmm... So Polkit should go into a different runlevel? 2022-04-15 02:14:37 you can try add a `need polkit` to libvirtd 2022-04-15 02:14:40 Or I need to setup the depends? 2022-04-15 02:14:43 ahh okay 2022-04-15 02:14:44 but usually 2022-04-15 02:14:47 you don't need that at all 2022-04-15 02:14:50 shrug 2022-04-15 02:15:08 I'm not using OpenRC in parallel so I'm surprised in this change in behavior 2022-04-15 02:15:21 Or does this go back to how it doesn't actually supervise processes and we need s6? 2022-04-15 02:16:20 i don't think any of this relates at all 2022-04-15 02:16:30 unless merely restarting libvirtd instantly fixes your issue 2022-04-15 02:16:47 I restarted polkit and then libvirtd and that seems to have fixed it 2022-04-15 02:17:08 hm 2022-04-15 02:17:12 weird then 2022-04-15 02:17:17 just the need should be fine i guess 2022-04-15 02:17:53 Can that be defined by default in our packaging, or is this not something folks commonly run into? 2022-04-15 02:18:10 a `want polkit` would be fine for that, but not sure if someone would want to add it 2022-04-15 02:19:40 Downsides? 2022-04-15 02:21:53 none if not installed 2022-04-15 02:21:59 otherwise it starts polkit if it's installed 2022-04-15 02:22:20 Is this the type of thing that would be welcome in an issue or an MR, or nah? 2022-04-15 02:22:37 sure, if you want 2022-04-15 02:22:40 but you should first do a test 2022-04-15 02:22:50 make sure neither are in rc-update (started default) 2022-04-15 02:22:54 both are stopped 2022-04-15 02:23:10 then start virtd first, then polkit, and ensure every time you try that, it's broken for a user 2022-04-15 02:23:14 and only in the opposite configuration it works 2022-04-15 02:23:20 Ahhh, gotcha 2022-04-15 02:23:24 Makes total sense 2022-04-15 02:23:40 that would mean you always need polkit to do something for it / be available already 2022-04-15 02:23:46 and then in that case, the want would make sense 2022-04-15 02:23:57 By the way, you got my rc-status | grep crashed to return empty with this fix, so 😍 2022-04-15 02:24:23 :) 2022-04-15 02:24:38 i don't think any of those crash even with the wrong ordering 2022-04-15 02:24:38 oooh! same for rc-status | grep stopped 2022-04-15 02:24:48 Usually polkit is stopped... it must depend on something else 2022-04-15 02:25:09 They were all returning crashed until I changed the runlevel of polkit to sysinit and the others to boot 2022-04-15 02:25:17 Three reboots so far, and all clear 2022-04-15 02:25:23 normally it starts automatically by dbus acrivation 2022-04-15 02:25:31 but for services i guess this doesn't work 2022-04-15 02:26:12 i.e. user things like pipewire would start, and start polkit+rtkit via dbus themselves, to request realtime persmissions 2022-04-15 02:26:22 no rc- used 2022-04-15 02:26:57 hmm, never had problems with that after I used wireplumber, so that sounds right 2022-04-15 02:30:36 2022-04-15 02:42:47 zxcva 2022-04-15 11:17:40 Hi, I can't enter special polish characters despite ,,pl'' keymap being set, although they are properly displayed and I can paste them. Thanks in advance. 2022-04-15 14:02:11 is there a way to configure the alpine iso to execute a script from an url? like a kernel command line argument like script=http://10.0.0.1/install.sh? 2022-04-15 14:03:15 not directly, but you can specify an overlay 2022-04-15 14:03:53 like from initrd? 2022-04-15 14:04:06 do you have an example? :-) 2022-04-15 14:05:17 I was looking at https://gitlab.alpinelinux.org/alpine/mkinitfs/-/blob/master/mkinitfs-bootparam.7.in but does not seem to have anything that I could use. 2022-04-15 14:05:24 https://gitlab.alpinelinux.org/alpine/mkinitfs/-/blob/master/initramfs-init.in#L613 2022-04-15 14:05:50 https://gitlab.alpinelinux.org/alpine/mkinitfs/-/blob/master/mkinitfs-bootparam.7.in#L109 2022-04-15 14:07:02 oh I completely missed it. 2022-04-15 14:07:04 thank you! 2022-04-15 14:36:29 hello, was anyone able to run flatpak's firefox with vaapi acceleration (amdgpu)? 2022-04-15 15:04:48 works for me without flatpak, didn't try flatpak though 2022-04-15 15:56:01 same 2022-04-15 18:45:21 Hi, I've got a problem regarding GNOME and X11. Whenever I try to start GNOME with startx (my .xinitrc: https://pastebin.com/S8u1zqD8) it's behaving like a powerpoint slide. I can switch between tty's and GNOME renders frame by frame. When I try using gdm, it just kicks me back to the login-screen. Mind you, wayland works perfectly fine. I am on an AMD RX 6800 XT and a Ryzen 5 2022-04-15 18:45:23 3600. Here's the latest Xorg.log: https://pastebin.com/vhABaUj2 2022-04-15 18:45:55 It'd nice if anyone could help me out! 2022-04-15 18:46:57 linuxuser: I never got GNOME working with anything but GDM/Wayland, so I'm surprised you have that even 2022-04-15 18:47:08 Do you need X11 or another Display Manager for some reason? 2022-04-15 18:48:03 Yeah, I'd like to play games at a high refresh rate without VSync 2022-04-15 18:48:35 Wayland adopts the refresh rate of my 60hz monitor, so my high refresh-rate one looks choppy 2022-04-15 18:49:48 Are you part of audio,video,input groups? 2022-04-15 18:50:25 I made sure I was beforehand! 2022-04-15 18:50:37 I am even part of the plugdev group 2022-04-15 18:50:44 Like recommended when install GNOME 2022-04-15 18:50:51 installing* 2022-04-15 18:51:07 [ 63.611] (WW) Open ACPI failed (/var/run/acpid.socket) (No such file or directory) 2022-04-15 18:51:13 [ 63.616] (EE) xf86OpenConsole: Cannot open virtual console 7 (Permission denied) 2022-04-15 18:51:24 Do you have acpi in your rc-update runlevels? 2022-04-15 18:51:35 And that looks like maybe you need to be part of the tty group? 2022-04-15 18:51:59 s/acpi/acpid/ 2022-04-15 18:51:59 Saijin_Naib meant to say: Do you have acpid in your rc-update runlevels? 2022-04-15 18:52:41 Hmm, no I am not part of the tty group nor have I acpi in my rc-update runlevels 2022-04-15 18:53:01 Gonna fix that quickly 2022-04-15 18:53:07 I'll be back in a minute 2022-04-15 18:54:48 * rc-update: acpid already installed in runlevel `default'; skipping 2022-04-15 18:54:53 I am now in the tty-group 2022-04-15 18:55:03 Let me try booting in to xorg 2022-04-15 18:55:08 Here's to hoping it helps at all 2022-04-15 18:58:08 Yeah it didn't work :( 2022-04-15 18:58:33 not sure if you even need xf86-video-amdgpu aside from just mesa-dri-gallium 2022-04-15 18:59:43 I've always installed the amdgpu driver in other distros. Just to be sure, you know 2022-04-15 19:02:04 I guess I'll have to not use my second monitor when gaming? Still, thanks for trying to help me. 2022-04-15 19:02:35 Sorry I couldn't help more. Never got that figured out myself 2022-04-15 19:03:41 I mean, I atleast don't have tearing, so that's a plus in my book! 2022-04-15 19:05:05 Wait, do I even have to setup-xorg-base when I'm using wayland? 2022-04-15 19:06:31 I'm guessing yes, because of Xwayland 2022-04-15 19:13:01 uuuuh. Great question. I did, and it didn't seem to break Wayland/GNOME/GDM 2022-04-15 19:16:55 Anyway, thanks again for trying to help. I'll be logging off. Have a great day, folk! 2022-04-16 17:22:55 Unknown AMD Radeon GPU ChipID=0x98e4 2022-04-16 17:23:01 any ideas ? 2022-04-16 17:23:06 my x wont start 2022-04-16 17:25:57 whats even funnier if I remove glamor-egl i get libEGL.so.1 exec format error ... 2022-04-16 20:07:09 are the "trying to overwrite" errors i'm getting here (https://ebc.li/hh4.txt) known? is this something serious or can i just ignore it? 2022-04-16 20:47:16 seems like an error with a subpackage of prosody also providing a file the main package provides, hence the failed to overwrite and the error state 2022-04-16 20:47:18 does apk fix sort it? 2022-04-16 20:48:37 If it's a transition (the file was first owned by the main package, but now by the subpackage, or visa-versa), then yes 2022-04-16 20:51:19 apk fix throws the same 3 errors 2022-04-16 20:55:09 package would need to be fixed I suppose 2022-04-16 21:25:55 that sounds like they are now part of prosody itself 2022-04-16 21:26:55 you can install specific modules from modules instead, but they probably need to be removed entirely if it's the case 2022-04-16 21:26:58 looks like it, dunno 2022-04-17 01:23:55 o/ 2022-04-17 17:41:28 i'm trying to run setup-alpine but running across the ssl issue of wrong date. rdate throws refused or failed errors. how can i get the right time onto the system so it can be installed? thank you 2022-04-17 17:42:00 opviel: Just using 'date' 2022-04-17 17:45:05 ikke: you mean set manually 2022-04-17 17:45:27 yes 2022-04-17 17:45:41 kthx 2022-04-17 17:53:04 opviel: physical PC or RPI or VM? If its a physical PC you could always go into the BIOS/UEFI and correct it 2022-04-17 18:02:26 VM 2022-04-17 18:02:45 the host's date is correct 2022-04-17 18:09:22 opviel: which hypervisor? doesn't the hypervisor provide the VM's RTC with the correct date/time? 2022-04-17 18:12:23 minimal: libvirt/qemu/kvm 2022-04-17 18:12:38 i'll dig into that side of the equation then 2022-04-17 18:12:43 have you incorrectly set rtc to localtime 2022-04-17 18:13:10 Hello71: if it's not in the domain or network config, it's default 2022-04-17 20:13:55 I am a meet popsicle 2022-04-17 21:18:27 opviel: depending on wether the emulated hardware time is correct, you can try playing with "hwclock" 2022-04-17 21:18:47 and with "date -s" 2022-04-17 21:20:06 satmd: the issue is that the aarch64 linux-lts kernel has rtc-efi as a loadable module rather than built into the kernel, so the RTC date/time was correct, the system date/time was not, solution was either "hwclock -s" or to use the linux-virt kernel 2022-04-17 21:20:23 ah 2022-04-17 21:20:38 which has rtc-efi builtin (and so the kernel sets the system clock using the RTC) 2022-04-17 21:21:13 so looks like either linux-lts for aarch64 should be changed to build rtc-efi into the kernel 2022-04-17 21:21:54 or else mkinitfs should add rtc-efi to the initram and the initramfs' init should load it if UEFI on a aarch64 machine with lts kernel 2022-04-17 21:40:57 So i'm trying to get the teleport binary to work on AlpineLinux, https://goteleport.com/docs/installation/ .... $ ξ‚° root@marina ξ‚° ~/teleport ξ‚° teleport Sun Apr 17 17:40:41 2022 2022-04-17 21:40:57 ! $ ξ‚° root@marina ξ‚° ~/teleport ξ‚° 2022-04-17 21:40:57 Error relocating /usr/local/bin/teleport: __res_init: symbol not found 2022-04-17 21:41:12 Weird, it's a golang program 2022-04-17 21:41:31 I did apk add gcompat 2022-04-17 21:46:02 Personally I'd try their build instructions as opposed to trying to run a binary compiled for some alien system 2022-04-17 21:50:11 it would be nice to get the binary to run since ansible could invoke/copy etc 2022-04-17 21:51:50 it's a client binary, i guess I'll try to build the whole thing 2022-04-17 22:00:45 @zcrayfish ..... worked 2022-04-17 22:00:54 Wonder if I should make a package 2022-04-18 04:29:23 Anyone using tigervnc with Alpine as a server? 2022-04-18 04:29:48 It is failing similarly to XRDP, though the service just straight crashes so I'm thinking I must be missing a dependency somehow 2022-04-18 04:30:41 member of input,audio,video groups, setup vncpasswd, setup the allowed users, allowed the machines in xhost to avoi MIT magic cookie issue, made sure the firewall ports were open in GUFW... At a loss here 2022-04-18 06:38:43 hello 2022-04-18 06:40:11 can you help me on xrdp ? 2022-04-18 12:54:47 Why do I get "vim: applet not found" when I try to run vim? 2022-04-18 12:54:52 As root. 2022-04-18 12:55:52 sounds like vim symlinked to busybox but it doesn't have vim program? 2022-04-18 12:56:11 all i can get without vim installed is 'command not found' 2022-04-18 12:56:21 that does indeed sound like a busybox symlink 2022-04-18 12:56:24 or `busybox vim` 2022-04-18 12:56:30 busybox only have vi 2022-04-18 12:56:42 if you run busybox vim it will show you applet not found 2022-04-18 12:56:46 yeah 2022-04-18 12:57:28 How do I fix it? 2022-04-18 12:57:59 if you don't have installed vim, you can install it and it should replace the symlink hopefully 2022-04-18 12:58:16 or you can run 'command -v vim' to see where it sits and remove it 2022-04-18 12:59:16 Hund: `which vim` 2022-04-18 12:59:35 which is not always installed 2022-04-18 12:59:56 Weird. I reinstalled vim and it fixed it. 2022-04-18 13:02:02 it probably replaced symlink or has higher priority based on PATH :) 2022-04-18 13:03:10 type -a vim should show you if there's others 2022-04-18 13:04:48 i forgot about `type`, how widespread it is across shells? 2022-04-18 13:05:21 type is posix 2022-04-18 13:05:25 -a.. perhaps not 2022-04-18 13:05:43 It works at least on ash, bash and zsh 2022-04-18 13:05:56 ash has it, zsh has it, bash has it.. 2022-04-18 13:05:57 yeah 2022-04-18 13:07:00 seems like -a is not in ash 2022-04-18 13:07:16 it is 2022-04-18 13:07:38 it doesn't show all executables with same name in PATH 2022-04-18 13:09:38 ah 2022-04-18 13:09:39 indeed not 2022-04-18 19:08:12 re: greeters not depending on xorg, I've decided to switch to greetd, since sddm is basically dead (not really but who has patience to wait 2 years for a bugfix) and apart very janky script trickery with stealing wayland-session script from sddm, greetd + gtkgreet works and successfully flashbanged me in completely dark room 2022-04-18 19:11:15 hi i am trying to install dwm and i get a error saying that there is no stdio.h libary 2022-04-18 19:12:03 milan: are you compiling it yourself? 2022-04-18 19:15:17 yes 2022-04-18 19:17:08 install musl-dev (or better, build-base) 2022-04-18 19:45:12 i get a error when compiling in Makefile with the line 50 tic -sx st.info 2022-04-18 19:46:24 https://pkgs.alpinelinux.org/contents?file=tic&path=&name=&branch=edge 2022-04-18 19:46:51 you can run `apk add cmd:tic` 2022-04-18 19:48:19 thanks 2022-04-18 20:17:03 now i get a error saying that xinit cant connect to xorg server 2022-04-18 20:23:00 milan: are you building on a headless machine? 2022-04-18 20:23:00 You can look into xorg logs to find clues 2022-04-18 20:27:44 it couldnt open ati and other modules 2022-04-18 20:29:02 What kind of graphics card do you have 2022-04-18 20:29:03 i also get a permison denied but if i run it in root i just get a black screen 2022-04-18 20:29:17 You need to be in the video and input groups 2022-04-18 20:40:57 or install udev 2022-04-18 20:41:06 and use startx instead of xinit 2022-04-18 20:49:43 i used startx 2022-04-18 22:08:10 hi there 2022-04-18 22:31:25 guys, am working on a script to autosuspend a server when there is no user connected. the thing is i need to use the who command (coreutils) to know what users are conencted to server, but who command has no output. what can i do? 2022-04-18 22:43:06 btw am using alpinelinux in diskless mode 2022-04-18 23:14:04 who --version 2022-04-19 00:33:59 pepe: utmps 2022-04-19 00:34:08 oh, gone 2022-04-19 03:44:18 hey all. I haven't used my machine in a while and it's getting stuck in the boot sequence trying to load the ram disk. is there any spot I should look to investigate the issue? 2022-04-19 03:46:49 if the last message you saw was loading initrd then it probably actually loaded it, and the kernel start after it is what failed 2022-04-19 03:46:57 can't tell you why, though, sadly 2022-04-19 03:50:54 hm okay.. I'll poke around in the boot loader.. I haven't touched the default setup but I'm guessing I upgraded and then maybe the upgrade left it in a bad state 2022-04-19 03:52:52 it's possible you upgraded without -a 2022-04-19 03:53:37 and ran into the 3.15 simpledrm issue where the upgrade without -a did not upgrade mkinitfs so you were left with no graphics module on the initrd, and so the display has no output at start, but does actually boot otherwise 2022-04-19 03:53:47 aside from that, no idea 2022-04-19 03:55:55 is there any way I can verify if that's the issue from a recovery stick? I am logged in but can't remember the chroot commands I need to run 2022-04-19 03:57:11 https://superuser.com/questions/111152/whats-the-proper-way-to-prepare-chroot-to-recover-a-broken-linux-installation the chroot from here 2022-04-19 03:57:24 set up networking on the usb, copy /etc/resolv.conf into the chroot 2022-04-19 03:57:38 chroot in, apk upgrade -Ua, then perhaps it is fixed by that 2022-04-19 04:01:46 got it! I'll give that a shot. thank you 2022-04-19 05:01:27 okay I updated everything and got an error about grub-probe: "failed to get canonical path for /dev/nvme0n1p3". this is from grub-2.06-r2.trigger 2022-04-19 05:04:18 ah 2022-04-19 05:05:08 there was a weird fix for this i think 2022-04-19 05:05:33 but i'm not sure i remember it 2022-04-19 05:07:30 I didn't see much online.. can I search irc logs somehow? hah 2022-04-19 05:08:50 these channels are logged, sure, but that keywork is not in this or the prior month https://irclogs.alpinelinux.org/%23alpine-linux-2022-04.log 2022-04-19 05:08:53 i forgot where i saw it 2022-04-19 05:14:16 not sure if it's relevant, but for some reason I had to mount the boot disk before I could mount the root disk. I didn't bother with the swap partition either 2022-04-19 05:14:51 did you mount /dev? 2022-04-19 05:16:09 oh no I didn't. will updating again rerun the failed trigger? 2022-04-19 05:16:18 apk fix 2022-04-19 05:16:53 maybe you should rbind /proc and /sys as well 2022-04-19 05:17:44 kk. will try it again. thanks! 2022-04-19 05:18:50 cool, now just seeing a warning about os-prober not running to check for other partitions. seems okay? 2022-04-19 05:19:12 completed successfully though. 2022-04-19 05:19:17 Yeah, is most likely harmless 2022-04-19 05:19:29 unless you expect to have multiple boot partitions 2022-04-19 05:21:35 rebooted and still stuck loading the kernel :/ 2022-04-19 05:28:19 Do you get any message? 2022-04-19 05:29:01 no, it just hangs on the ramdisk message 2022-04-19 05:38:53 is there anything I can do to regenerate my grub setup? 2022-04-19 05:48:43 hmm... I have a /etc/default/grub.apk-new file... could that be related? 2022-04-19 05:59:16 eh didn't do anything. I'll give it a break for now and try tomorrow. thanks for the help! 2022-04-19 15:04:01 cog 2022-04-19 15:04:10 sorry wrong channel 2022-04-19 15:04:19 or rather, wrong window 2022-04-19 15:04:38 ACTION meant to type that into terminal 2022-04-19 18:06:13 hi i have installed dwm following the alpine wiki and when i try to start it with startx i get a error 2022-04-19 18:07:10 failed to open virutal consle permission denied 2022-04-19 18:42:37 are you part of input and video groups? 2022-04-19 20:02:15 what packages contains linux/soundcard.h 2022-04-19 20:02:49 https://pkgs.alpinelinux.org/contents?file=soundcard.h&path=*%2Flinux&name=&branch=edge&arch=x86_64 2022-04-20 03:47:13 Hello all 2022-04-20 03:51:57 Guys, some one knows how can i see how many users are logged in, I've tried with the who command but doesn't work 2022-04-20 03:55:36 utmp is broken so it won't work 2022-04-20 03:57:48 what other tool can i use? 2022-04-20 04:04:18 have you tried ps 2022-04-20 04:59:45 Aside from ps, if it's all terminal logins, I wonder if one could simply look in /dev/pts. 2022-04-20 09:05:37 broken as in "because of " or broken as in "fundamentally"? 2022-04-20 11:24:58 originally there was no utmp, then it got added via utmps (that you have to set up manually and reboot), but there is some bug in it (or so i was told) where the hardcoded path to wtmp socket is /dev/null so it fails 2022-04-20 11:25:19 the coreutils who should work with utmps set up, but not the others, or something 2022-04-20 11:25:19 shrug 2022-04-20 11:25:30 i never use these as they are essentially useless so i never gave it much testing 2022-04-20 11:25:48 there is some stuff that minimal is working on but idk 2022-04-20 11:25:50 so.. bug# unreported 2022-04-20 11:25:51 yea 2022-04-20 11:26:39 you could probably look at everyones else processes and look for login shell or ssh :P 2022-04-20 15:55:05 Hi, trying to get some help as I am stuck since last week on a RAID issue. I set up a cloud based OVH bare metal using IPMI and chroot to install Alpine instead of proposed Linux version. Everything is working ok but when I reboot I am stuck on boot process. I am able to get the error : /dev/md0 on /sysroot failed : No such file or directory. I checked my mdadm, my modules and 2022-04-20 15:55:06 added raid , did a mkinitfs, fstab is ok, UUID as well. Just before this error i got a md/raid1:md0 : active so it seems raid module is loaded and md0 active. I also tried to add to openrc boot mdadm and mdadm-raid but no more success. Any clue? 2022-04-20 15:56:17 blt: right after you get that error, can you check if /dev/md0 exists 2022-04-20 15:57:47 I wish I could but IPMI is not responsive after the issue and stucks at "Manufacturer : American Megatrends Inc." nothing can be done appart of rebooting 2022-04-20 16:00:41 another strange thing is that rsyslog or openrc are not producin 2022-04-20 16:00:53 g any log files in /var/log 2022-04-20 16:40:06 hitting an odd issue when trying to add my user to the `abuild` group. I've got an alpine VM (created via Lima/qemu on macOS), and I'm connecting over SSH. When I add my user to the group, it succeeds, and the user is an entry for abuild in /etc/group, and `groups $USER` shows the new group, but when I run `groups`, the new group isn't there, even after logging out and logging back in. When I try to run `build -r` it 2022-04-20 16:40:06 complains I'm not in the `abuild group` 2022-04-20 16:42:16 logging out and back in should give it 2022-04-20 16:42:21 not seeing it means you didn't actually relog 2022-04-20 16:42:38 as another test, you can `su - youruser` and it should also then appear 2022-04-20 16:42:39 yeah, that's my suspicion, perhaps rebooting the VM will do it. not sure what's up. 2022-04-20 16:44:21 well, rebooting the VM worked :| I wonder if lima does something weird with ssh 2022-04-20 16:52:04 shrug, it would just mean they persist the login somehow 2022-04-20 16:52:53 which would also mean if you `exec su - youruser` that change should also persist if it's kept :p (maybe) 2022-04-20 16:54:08 yeah, I didn't think about using su to check 2022-04-20 16:54:37 ok, so now that abuild works, having a new issue, where do I find `openssl1.1-compat-dev`? it's in the linux-lts makedepends, but I can't find it 2022-04-20 16:56:29 openssl-dev 2022-04-20 16:57:10 it was just a compat name added to openssl1.1-dev or whatever when we tried an openssl3 migration for 3.15 2022-04-20 16:57:18 then stuck around since we kept the changes for 1.1 packages 2022-04-20 16:57:29 we dropped the thing eventually, but the 1.1 naming stuck around in some places 2022-04-20 16:57:35 just openssl-dev provides it 2022-04-20 16:59:19 ikke: by the way I went to grub rescue / shell and I am able to browse the files , not 100% sure it is md0 as the naming are not the same and because it is raid1 maybe the files are on the partition anyway but that's another thing where I have been able to browse my / .. 2022-04-20 17:01:51 @psykose gotcha, that seems to work, I guess linux-lts requires updating, or the branch I'm on needs to be rebased :D 2022-04-20 17:02:32 which branch 2022-04-20 17:06:21 openssl1.1 still provides openssl1.1-compat-dev 2022-04-20 17:06:27 so I don't see any issue 2022-04-20 17:06:57 I'm on https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/28218 trying to build it locally, I rebased it a week or so ago, but haven't since 2022-04-20 17:08:23 adding compat package should still work 2022-04-20 17:09:39 so should `apk add openssl1.1-compat-dev` work? 2022-04-20 17:09:57 yes 2022-04-20 17:10:30 it does not work on 3.14.3 2022-04-20 17:10:45 `openssl1.1-compat-dev (no such package):` 2022-04-20 17:11:08 if you have 3.14 repos it won't, because it only exists on 3.15+ 2022-04-20 17:11:23 you are building master/3.15 aports with 3.14 /etc/apk/repositories 2022-04-20 17:11:48 you can `apk add -r openssl1.1-compat-dev openssl-dev` to force it to still work 2022-04-20 17:11:50 er 2022-04-20 17:11:52 -t not -r 2022-04-20 17:12:00 so you don't have to edit the files 2022-04-20 17:12:02 if it's your wish 2022-04-20 17:12:13 ah, so to build packages I should be on 3.15 2022-04-20 17:12:23 well, it depends what you are aiming for 2022-04-20 17:12:31 shoot, I probably need to build the 3.14 kernel then, didn't think about that 2022-04-20 17:13:07 MR is pointed at master branch, so it builds with whatever changes are currently in master/edge branch 2022-04-20 17:13:23 I'm trying to get debug symbols so I can run pahole on them to enrich them with some BTF metadata for BPF process traces 2022-04-20 17:13:30 ah 2022-04-20 17:13:35 right 2022-04-20 17:13:38 and this PR is just the one that seems to be closest to helping me with those goals 2022-04-20 17:13:39 don't remember how to add those 2022-04-20 17:13:50 it's easy to add to the kernel, it's a CONFIG_ var 2022-04-20 17:13:55 if you will be mring it everything will be against edge anyway 2022-04-20 17:14:11 you can switch to 3.14-stable branch and build it from there 2022-04-20 17:14:18 yeah 2022-04-20 17:14:44 would need to apply the changes from MR though if that's required 2022-04-20 17:14:59 I know I can just build the kernel itself with everything I need + get debug symbols, but that's not super useful for me atm since I can't really supply my own kernel with rancher-desktop for the real place I'm using this stuff 2022-04-20 17:15:11 yeah, ill rebase on 3.14-stable, shouldn't be hard. 2022-04-20 17:15:55 now I just need to figure out wtf Lima is doing such that I can't install kernel/firmware packages (seems like they've got a readonly loopback/squashfs setup for kernel/modules/etc) 2022-04-20 17:49:22 that's alpine 2022-04-20 17:49:38 yeah, I just discovered the details on it 2022-04-20 17:51:15 but I'm still trying to figure out if that's something that can be changed 2022-04-20 17:53:19 basically trying to make it so I can install the kernel packages since atm it fails on the RO filesystem 2022-04-20 18:09:51 anyone have any advice for how to get through that issue? I can't build the kernel package because it fails on builddeps because it fails on installing linux-firmware due to the modloop setup 2022-04-20 18:10:46 apk add linux-firmware-none 2022-04-20 18:10:53 but that will probably fail too, not sure 2022-04-20 18:12:16 that works, but I'm not sure how that fixes the ability to build the package? 2022-04-20 18:12:25 it doesn't need to pull in linux-firmware anymore 2022-04-20 18:12:50 ah, well, the build is progressing onto compiling now, so I suppose that did it 2022-04-20 18:13:33 I can't help but feel like "Im doing it all wrong" but I don't have extra hardware for a non-VM setup so idk 🀷 2022-04-20 18:14:31 no, you're doing it all right 2022-04-20 18:14:48 though it would be a bit easier if it was a sys install and not one of the others :) 2022-04-20 18:19:50 yeah, definitely agree on that point. 2022-04-20 20:38:48 suppose I did want to install a custom build of my kernel, I think it's just an `apk add` in theory, but with mod loop I'm guessing I'm gonna hit an issue 2022-04-20 20:45:37 You'd need to generate a modloop as well 2022-04-20 20:54:11 probably easiest to build my own ISO with mkimage and such at this point now that I've sorted out the intermediate steps 2022-04-20 20:54:51 lima has a repo for doing all this via docker I think too, so probably not too hard 2022-04-20 22:32:51 having trouble sorting out how I actually get the ISO to be built with my custom image at this point. 2022-04-20 22:35:15 I see I can set kernel_flavors in the profile, but not sure how that relates. 2022-04-20 22:36:37 maybe I just need my package to be available by modifying my /etc/apk/repositories and it will use it, assuming the flavor is the same 2022-04-21 05:49:16 so im not sure if there's any clear documentation on this already but: alpine for x86 is built targeting an i586 CHOST. is it considered a bug if a package does not work on a machine meeting that target and can i send patches to fix it 2022-04-21 05:52:32 e.g. mesa actually effectively targets *i786* (requires sse2, aka pentium4) as a minimum for x86 regardless of your actual chost unless you explicitly pass an option 2022-04-21 05:52:50 and alpine doesn't pass that option, so none of my x86 machines can run alpine and have working mesa 2022-04-21 05:53:35 chost, cbuild, whatever 2022-04-21 05:54:07 linearcannon: https://gitlab.alpinelinux.org/alpine/tsc/-/issues/20 2022-04-21 05:54:39 https://gitlab.alpinelinux.org/alpine/tsc/-/blob/master/minutes/2022-01-11.md#user-content-baseline-for-x86-issue-20 2022-04-21 06:00:57 hmm. i'm not entirely sure if that answers my question, but ok 2022-04-21 06:01:27 like. i'm happy to bear the burden of testing my own use case and fixing all the problems, and i can resolve the issues of mesa and rust and such needing sse2 2022-04-21 06:02:14 i guess i'll assume that conclusion means i should send patches 2022-04-21 06:09:36 linearcannon: It basically means, we try to support i586 when possible 2022-04-21 06:11:40 so if it is possible for a package to support i586 but it currently doesn't, i should send a patch 2022-04-21 06:12:07 i haven't heard of mesa needing sse2. i would personally (cannot speak for tsc) be somewhat opposed to disabling sse2 on rust 2022-04-21 06:12:24 https://gitlab.alpinelinux.org/alpine/tsc/-/issues/35 2022-04-21 06:13:45 my thing on that is, rust components are now necessary to build a fully working graphical system, which is what i am trying to do 2022-04-21 06:14:00 now, if you could patch it to generate both x87 and sse2 code with automatic target selection, that would solve that problem in theory, but then i would be suspicious that non-upstream code of that nature could be reliable 2022-04-21 06:14:01 when targeting i586-unknown-linux-gnu rust already does disable sse2, but not for i686 (even though sse2 is i786) 2022-04-21 06:14:02 mhm, gnome 2022-04-21 06:14:40 there's also some bugs for i586 target, though gentoo carries patches for those 2022-04-21 06:14:52 and the bug has an issue upstream 2022-04-21 06:16:16 at some point your hardware is strange enough that it doesn't make sense to be supported by binary distro. imo non-sse2 x86 cpus have reached that point; i586 is supported as "best effort", not a serious target 2022-04-21 06:16:55 now, if you are willing to manually compile packages for your hardware, so https://gitlab.alpinelinux.org/alpine/tsc/-/issues/35 doesn't apply, then i think it might be ok 2022-04-21 06:17:09 i'm not really sure what the use case is for most people running i686 alpine, but the majority of cpus that support sse2 also support x86_64, with the exception of early pentium4 chips, and the mobile chips from around and slightly after that era 2022-04-21 06:17:14 *i586 2022-04-21 06:18:05 afaik nobody else is quite sure either 2022-04-21 06:18:21 non-64-bit atom is up till 2018 or something ridiculous though 2022-04-21 06:19:38 ... iirc... 2022-04-21 06:20:59 so like. if most sse2 cpus can run x86_64 version anyway im not sure why sse2 would be targeted for x86_32 and i dont think it would make sense for alpine to bump to i786 (or "i686+sse2" which is the same thing). i also dont understand why upstream projects are doing that 2022-04-21 06:21:44 ah, the 32-bit-only cpus were stopped in like 2010, but crappy oems bought cheap mobo and bios and disabled 64-bit operation 2022-04-21 06:25:29 i guess you don't know: sse2 is mandatory for correct 64-bit double implementation. without sse2 you must use x87 which only has weird float sizes, or software emulation 2022-04-21 06:25:58 x87 is slow, but not terribly so compared to those old cpus. the bigger problem is that the floating point calculations result in different values 2022-04-21 06:29:09 ...im pretty sure thats not true. i'd have to check but i'm certain there's a flag that forces it to follow ieee rounding rules for 64-bit (i write hand-optimized x87 code for i486 in my spare time, but i still wouldn't call myself an expert) 2022-04-21 06:30:33 if you dont set the flag then yes, it will be slightly off compared to ieee754 because it has higher precision 2022-04-21 06:31:11 80-bit rather than 64 2022-04-21 06:38:46 wait. x87 64-bit double uses a 53 bit mantissa, not 52. i didn't realize that. yeah, that is wrong. 2022-04-21 06:43:22 there are various flags for this purpose but according to (my interpretation of) dalias, none of them allow x87 to work precisely like binary64 because such a mode does not exist and cannot be efficiently emulated 2022-04-21 06:43:56 and yes, i believe the issue with the fpu flag is that x87 64-bit has too much mantissa (and insufficient exponent) 2022-04-21 06:44:44 yeah, that's correct based on my understanding, i did not realize the bit format was different for 64-bit float. though it could be emulated rather quickly if you were hand-writing assembly, or had support in the compiler, but that is a lot 2022-04-21 06:45:03 very large amount of effort for supporting what i will acknowledge is a rather old platform 2022-04-21 06:45:38 still, what i care about (mainly rust, mesa) doesn't have this problem that php does 2022-04-21 06:45:47 i'm not familiar enough with floating-point math to say for sure, so i defer to dalias's claim that it is not possible to emulate efficiently for all operations, even with compiler support 2022-04-21 06:46:02 well rust does, no? 2022-04-21 06:46:42 specifically, dalias said https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/28340#note_200036: Emulating conforming double arithmetic on top of excess precision is actually very hard. I'm not sure if there's any good way to "patch it up". x87 provides a flag indicating the direction of the last rounding which may help, but the only obviously-correct approach I know 2022-04-21 06:46:44 is just ignoring the fpu and doing soft-float for everything. 2022-04-21 06:48:39 i agree that part of the issue is that the user base for alpine x86-32 is not clear, and whether they (or prospective future users) care more about wide hardware support or exact floating-point consistency 2022-04-21 06:49:10 it's possible, but you have to do manual conversions to and from 80bit when loading or storing to memory. there's a fast trick for this. you'd also have to do manual rounding with each operation. it would still be about the same amount of work as writing a softfloat implementation (but would be faster) 2022-04-21 06:49:15 (the third option being to make everything automatically dispatch between x87 and sse2, but i think we agree that's not going to happen) 2022-04-21 06:49:43 yeah 2022-04-21 06:50:02 it's actually pretty easy with gcc/clang but rust offers no functionality for this 2022-04-21 06:50:39 i don't know what dalias meant by that exactly. i suspect he may have been referring to operations beyond the basic set of +-*/ 2022-04-21 06:51:00 well, in that issue, -fexcess-precision was my initial proposal, before i found that it does not operate as the name implies 2022-04-21 06:52:57 i meant the automatic dispatch not fixing floats, that's hard 2022-04-21 06:53:10 little known feature, but it's there 2022-04-21 06:53:20 i think -ffloat-store -fexcess-precision=standard probably does most of the work 2022-04-21 06:53:33 ifunc doesn't work on musl 2022-04-21 06:53:52 ah, it doesn't? well that's a shame 2022-04-21 06:55:41 anyway rust as well as most rust programs i've cared to test work down to i486 just fine, it's just that its i686 defaults to having sse2. and currently if unpatched it also builds some tiger lake instructions in that are NOPs on i686 but invalid on i586 and below. 2022-04-21 06:55:51 i think https://www.openwall.com/lists/musl/2014/11/11/2 is relevant (about ifunc) 2022-04-21 06:57:07 afaik everyone agrees that the idea of target_clones is good, but ifunc is overly powerful and underspecified 2022-04-21 06:57:28 you mean cet? 2022-04-21 06:57:40 yeah that one 2022-04-21 06:57:56 i believe i brought up cet in one of those tsc issues 2022-04-21 06:59:14 i agree that cet on x86-32 is a bit silly, because anybody using cet-capable cpus is almost certainly using x86-64, or running in a non-security-critical environment (e.g. ci/cd) 2022-04-21 06:59:45 but intel has never let practicality stand in the way of a new isa extension 2022-04-21 07:00:43 anyway i have to get to bed 2022-04-21 07:01:07 nn 2022-04-21 07:01:13 i guess tomorrow i'll write a patch for mesa 2022-04-21 07:01:15 lol 2022-04-21 07:02:00 i wouldn't be terribly surprised if mesa tests somehow depend on ieee 754 binary64 2022-04-21 07:02:45 i am about 100% sure i've built mesa on i486 with passing tests 2022-04-21 07:03:39 big endian, on the other hand, is another story. and i've got a bunch of big endian stuff lying around too 2022-04-21 07:03:54 i've also been trying to fix that, but every missing byteswap i find uncovers two more 2022-04-21 07:04:00 i don't think i brought up in the discussion so far that even if none of the users care about exact floating-point results, many of the tests care, and alpine devs don't really want to spend ages chasing down those test failures 2022-04-21 07:04:38 from my view it seems that big endian is quite clearly dead in the long term 2022-04-21 07:04:49 yeah. i'm sad about that. i'm trying to keep it going as long as i can though 2022-04-21 07:06:45 it looks like alpine's rust is actually built for i586 properly and so works on i686 without sse2, so at least i don't have to open that can of worms 2022-04-21 07:07:03 anyway i actually should go bye 2022-04-21 07:10:35 are you sure? i don't think alpine rust is patched to remove sse2, and afaik the issue is still present upstream 2022-04-21 07:11:02 i reckon if you just do rustc --version it doesn't need any floating-point calculations 2022-04-21 08:07:21 Hello71: upstream rust does not generate sse2 if the target is i586. it doesn't need patching for this, it only needs patching if targeting i686 without sse2. i can tell you this because i have rust running on i586 netbsd, running on an i486 with a kernel patch i wrote that emulates i586 opcodes 2022-04-21 08:07:34 and i also just tested on alpine on a pre-sse2 machine, building a rust program that uses f64, and it worked fine 2022-04-21 09:49:17 Hi, I would like to know if there is a way to configure a boot splash screen in Alpine Linux? 2022-04-21 10:33:24 Sprinter: that depends on what boot loader you use, if you have GRUB you can follow it's instructions on how to change splash screen 2022-04-21 10:53:48 linearcannon: if there is a flag to pass for the x86 mesa build to make it i586 compliant then that would be fine to add, and you should open an mr/send a patch/whatever 2022-04-21 10:54:13 i'll merge it if you say it does what it's supposed to 2022-04-21 10:54:22 i586 is indeed best effort, and so this would be the effort :p 2022-04-21 10:55:07 though note that mesa 22 removed all the dri-classic drivers and there is only gallium left.. and the existing mr i have for 22 just leaves that void 2022-04-21 10:55:24 afaik there is an amber tree now of the old 21.x line for the classic drivers, and i saw arch package it separately 2022-04-21 10:55:41 i should probably also get a separate mesa-amber thing done before 22, and before the 3.16 release 2022-04-21 10:56:02 if you have the actual hardware to test it, you might want to do that yourself instead, i wouldn't mind a mesa-amber aport of the classic drivers 2022-04-21 10:57:09 (and you are definitely more the expert on all that shit than me, so it would be appreciated, if you want to help with the support) 2022-04-21 15:08:14 im a bit confused reading https://wiki.alpinelinux.org/wiki/Custom_Kernel and trying to figure out the correct method to add new CONFIG_ options to the kernel package. in particular, `CONFIG_DEBUG_INFO_BTF=yes` 2022-04-21 15:10:07 kernel sources are in src/linux-x.y.z afaik 2022-04-21 15:10:55 you can do make menu config and look for relevant option there 2022-04-21 15:13:27 if you are modifying the existing ones you can add it into the config files you see in the folder 2022-04-21 15:13:43 if you are modifying a kernel build tree you can use menuconfig and do stuff with the resulting .config 2022-04-21 15:13:50 chancez: kernel is in aports, you can edit the relevant config files as desired 2022-04-21 15:13:52 you can also ask ncopa to add the option for you next update 2022-04-21 15:14:17 e.g. https://git.alpinelinux.org/aports/tree/main/linux-lts?h=3.15-stable 2022-04-21 15:14:25 remember to `abuild checksum` after you update config files 2022-04-21 15:15:04 yeah, so tried that, but hit a snag, basically I added the config flag, ran `abuild -r` and it prompted me to set some new configs for arm64, and then I cancelled after the first set of new config options, and checked the generated .config, but its was missing the config option I manually added. 2022-04-21 15:15:09 but .config.old had it 2022-04-21 15:15:27 config flag? 2022-04-21 15:15:44 it adds it into the new .config in src/build-*/.config 2022-04-21 15:15:49 sorry, terminology is escaping me 2022-04-21 15:15:50 not into the same folder you are already in 2022-04-21 15:15:54 I did this: `find . -name 'config-*' -type f -exec bash -c 'echo "CONFIG_DEBUG_INFO_BTF=yes" >> $0' {} \;` 2022-04-21 15:16:02 ncopa: that being said, before I forget, please s/CONFIG_VIRTIO_CONSOLE=m/CONFIG_VIRTIO_CONSOLE=y/ for the virt kernel; else you don't get early console for qemu microvm 2022-04-21 15:16:32 chancez: you did $0 in your script 2022-04-21 15:16:33 then ran `abuild -r`, got prompted before some options, held down enter for a bit, then used ctrl-c. `cd src/build-lts.aarch64` and `diff .config .config.old` 2022-04-21 15:16:39 so it would exec append it to uh 2022-04-21 15:16:42 not the file you're looking for :p 2022-04-21 15:16:49 nah, it worked, find exec is...odd 2022-04-21 15:17:01 I forget why but it's $0 instead of $1 :D 2022-04-21 15:17:22 also it needs to be y 2022-04-21 15:17:24 not yes 2022-04-21 15:17:26 which is probably why you got what you got 2022-04-21 15:17:31 ah that might be it 2022-04-21 15:18:02 personally I'd opt for a more careful way of editing the configs :p 2022-04-21 15:18:10 I'm open to a suggestion 2022-04-21 15:18:12 now your config may have duplicate lines for CONFIG_DEBUG_INFO_BTF 2022-04-21 15:18:26 it doesn't, I checked, nothing sets it 2022-04-21 15:18:42 that's odd 2022-04-21 15:18:58 ACTION fwiw I'm rebased on stable-3.14 since my VM is 3.14 2022-04-21 15:19:08 ah wait 2022-04-21 15:19:16 the kernels don't set CONFIG_DEBUG_INFO=y to begin with 2022-04-21 15:19:21 are you sure a sub-option of it will work then :P 2022-04-21 15:19:29 (i recommend menuconfig for this exact kinda reason) 2022-04-21 15:19:42 (or even better, nconfig) 2022-04-21 15:19:51 no not really, I'm basically just trial/error here, only just built the kernel without any changes besides the -dbg package 2022-04-21 15:20:06 I could use menuconfig, just gotta find the right spot 2022-04-21 15:20:07 right, well, kernel options introduce and rely o ndependencies 2022-04-21 15:20:15 so just appending one to the .config may not result in what you want 2022-04-21 15:20:49 menuconfig/nconfig have a built-in search feature to locate the menu by CONFIG_ name 2022-04-21 15:21:04 from the top of my head it was f8, f9, or f10, but it's been forever, and it tells you at the bottom of the screen anyway 2022-04-21 15:21:27 adding debug_info (-g) will require adding some form of -dbg subpackage too 2022-04-21 15:21:34 you should really just ask ncopa for it :) 2022-04-21 15:21:41 but yeah the documentation is a bit sparse on that wiki 2022-04-21 15:21:59 yeah I'm totally going to be asking for it, but I also want it locally to test some stuff *soon 2022-04-21 15:22:22 basically, I took this: https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/28218/ and stripped it of the non-dbg changes, and rebased on 3.14-stable 2022-04-21 15:22:45 Shiz: like this? c3dcc58666c4246164bce1584423a98401fbefd6 2022-04-21 15:22:48 and then built it, and verified the dbg package produced the package as I expected, at least I thought. 2022-04-21 15:23:03 now I want to build it, but with the BTF metadata 2022-04-21 15:23:24 chancez: generally what I'd do (instead of on the wiki) is get a clean abuild env again, then run `abuild prepare` so everything is unpacked and into place, then cd build-virt.x86_64 and run make nconfig 2022-04-21 15:23:30 (or whatever is applicable for your kernel) 2022-04-21 15:23:39 then in that dir you have a new .config you can copy back to config-virt.x86_64 2022-04-21 15:23:42 if you just want local testing, then make the config only for what you will be using 2022-04-21 15:23:43 that sounds exactly like what I want 2022-04-21 15:23:58 you can copy the config-lts into src/kernel-tree/.config, cd to it, use menuconfig/nconfig 2022-04-21 15:24:00 then just copy it back 2022-04-21 15:24:13 psykose: abuild prepare gives you nice build folders to use for that instead of the kernel tree :) 2022-04-21 15:24:33 it's the same process, you are still going into it and running `make menuconfig` 2022-04-21 15:24:49 ah this is amazing, no more ctrl-c 2022-04-21 15:25:08 ncopa: nice, did i complain about this earlier 2022-04-21 15:25:16 because i was seting this stuff up exactly last week/two weeks ago 2022-04-21 15:25:18 :p 2022-04-21 15:25:18 i think someone else asked for it 2022-04-21 15:25:21 unless that was also you 2022-04-21 15:25:27 regarding booting a vm from macos 2022-04-21 15:25:34 also needed hvc0 2022-04-21 15:25:36 yeah 2022-04-21 15:25:38 nah, i just use vmware on macos :p 2022-04-21 15:25:58 but cool, nice 2022-04-21 15:26:08 then i can drop my mkinitfs patch too 2022-04-21 15:26:08 its only in edge so far 2022-04-21 15:26:15 could it be backported to stable? 2022-04-21 15:26:19 it would help for my vm deployments 2022-04-21 15:26:37 possibly. we will ship 3.16 in a month or so 2022-04-21 15:27:02 sure, but i'm not sure if i can major-bump my vms easily like that :p 2022-04-21 15:27:13 right 2022-04-21 15:27:31 i might have a look at it, but not today 2022-04-21 15:27:36 ofc ofc 2022-04-21 15:27:37 that's why I'm trying to figure this out lol, otherwise id just ask someone else :D 2022-04-21 15:27:54 Shiz: would be nice if you coudl file an issue about it, so I dont forget it 2022-04-21 15:28:30 sure, i may just throw in a pr :p 2022-04-21 15:28:34 depending on my time 2022-04-21 15:28:42 also long time no speak, hope you are well, and stuff 2022-04-21 15:31:09 so I ran make prepare, then used `/` to find the BTF option, enabled it + debug symbols, but now it seems like the other menu options that were added when I ran `make prepare` aren't in there 2022-04-21 15:32:07 actually, maybe not, was diffing .config and .config.old 2022-04-21 15:32:36 :P 2022-04-21 15:33:14 you don't have to diff anything, just copy it over the old one 2022-04-21 15:33:21 and then `git diff` to verify what changed in the tree 2022-04-21 15:33:30 and if you fuck anything up git is still there 2022-04-21 15:34:16 yeah, was just trying to compare what it added 2022-04-21 15:39:19 geez I have no idea about why some options are gone now but, hopefully they're not important 2022-04-21 15:41:50 if in doubt, post diff 2022-04-21 15:42:17 don't go off hope, look them up in lkddb :p 2022-04-21 15:42:20 I already pushed it up to my branch, one sec :D 2022-04-21 15:42:31 the Diff is pretty big, hence "welp..." 2022-04-21 15:42:48 https://github.com/chancez/aports/commits/linux_dbg_3.14 most recent commit `main/linux-lts: Add CONFIG_DEBUG_INFO_BTF=y` 2022-04-21 15:43:07 ah crap and I need to update the LTS .config too (only did -virt) 2022-04-21 15:45:37 uhhh 2022-04-21 15:45:43 that looks like you kinda reset most of the config 2022-04-21 15:45:50 yeah 2022-04-21 15:45:57 probably by taking a defconfig, and adding the option 2022-04-21 15:46:02 instead of starting from the existing one 2022-04-21 15:47:12 hmm I must have missed a step 2022-04-21 15:47:15 did I need to `cp config-dirt.aarch64 src/build-virt.aarch64/.config` ? 2022-04-21 15:47:34 I ran prepare, then `cd src/build-virt.aarch64` and then `make menuconfig` 2022-04-21 15:48:30 if in doubt, just do it in the src/linux-5.10 tree :p 2022-04-21 15:49:13 I really just need to re-learn the `make *config*` options :| 2022-04-21 15:49:40 if you don't run something like defconfig or whatever to reset it back to default it should be fine 2022-04-21 15:50:05 prepare seems to run `make listnewconfig oldconfig`, hence me seeing the new options, now that makes sense. 2022-04-21 15:50:35 yes, and oldconfig prompts you to pick the new ones 2022-04-21 15:50:44 (olddefconfig would instead just pick the defaults for all of them) 2022-04-21 15:53:05 btw which option is for present hardware only? 2022-04-21 15:53:54 make help 2022-04-21 15:54:04 :D 2022-04-21 15:59:13 @psykose shell history says I copied the LTS config into the virt config last time, hence the diff :D 2022-04-21 16:01:25 :) 2022-04-21 16:01:34 so now I just `abuild -r` right? 2022-04-21 16:01:39 probably, sure 2022-04-21 16:01:45 and ideally it won't prompt me again since I already confirmed the new settings 2022-04-21 16:02:46 well it fails now, which is good since it's failing on python for some bpf stuff 2022-04-21 16:03:56 it needs python at build time for it? 2022-04-21 16:03:58 weird 2022-04-21 16:05:38 yeah, I think for BTF it'll also need pahole which will be fun since it seems like that relies on libbpf from edge 2022-04-21 16:05:55 and I think pahole itself is only in edge 2022-04-21 16:07:01 yes 2022-04-21 16:07:15 uhh if these things are needed at kernel build time then it doesn't sound easy to add 2022-04-21 16:08:00 pahole is in testing/ for one, and needing python built before the kernel is.. well, not really sure about that one 2022-04-21 16:08:24 normally i would say these sorts of things should just be thrown into linux-edge instead, but linux-edge is even more conservative than -lts because of who maintains it 2022-04-21 16:17:51 yeah, I'm not super sure how hard getting any of this into upstream is going to be 2022-04-21 16:18:06 hence mostly focusing on local use atm 2022-04-21 16:27:42 @psykose where do I even find linux-edge? 2022-04-21 16:27:52 community/ and only from 3.15 2022-04-21 16:27:55 wait, I realized, you said testing, and I'm in `main`, duh 2022-04-21 16:28:01 well 2022-04-21 16:28:25 to even make a change with this you would have to move pahole to main/ or whatever 2022-04-21 16:28:37 (for the mr, not for doing it yourself) 2022-04-21 16:28:42 yeah 2022-04-21 16:28:42 the way you're doing it now is fine for local use 2022-04-21 16:29:31 yeah I'm curious what would be needed to get all this for say,...3.16? :D 2022-04-21 16:29:44 if not 3.16 then 3.17 2022-04-21 16:30:12 writing out the full list of things needed for it and showing the config options and then also doing the -dbg split on top of it and ncopa having the time to look at it 2022-04-21 16:30:23 I think for earlier than that, I can add BTF metadata as long as the debug symbols/package exists 2022-04-21 16:30:28 and that goes for getting it done ever; whether it's in time for 3.16 is just whether he sees and approves etc 2022-04-21 16:30:59 gotcha, so what I'm doing now for LTS, but basically just creating an issue & documenting what I found, roughly 2022-04-21 16:31:06 you should really open an issue and describe the stuff in there 2022-04-21 16:31:12 yep 2022-04-21 16:36:43 looks like an issue already exists https://gitlab.alpinelinux.org/alpine/aports/-/issues/12563 2022-04-21 16:37:37 ah, and seemingly forgotten about 2022-04-21 16:37:38 haha 2022-04-21 16:37:58 was hoping Martinetd would be in here 😒 2022-04-21 16:41:34 as far as I could tell though, pahole w/ libbpf from edge seems to be working 2022-04-21 16:41:51 that sounds pog 2022-04-21 16:41:53 but haven't been able to fully exercise it on the kernel since I haven't built one with debug symbols yet. 2022-04-21 16:42:07 but it doesn't fail on --help at least, which it does if you use libbpf from main 2022-04-21 16:42:23 we'll see in about an hour if it works 2022-04-21 17:25:22 woot I think it worked 2022-04-21 17:25:51 pahole works on the vmlinux, the question is, do I even need to run pahole, since I think the CONFIG option should do that. 2022-04-21 17:28:45 seems like `llvm-objcopy --dump-section .BTF=.btf.vmlinux.bin /usr/lib/debug/5.10.109-1-virt/vmlinux` works 2022-04-21 18:17:36 so im struggling to install these packages in a different host because of the signature, what do I need to do to trust the pub key I used to sign my packages with? 2022-04-21 18:17:48 I tried copying it to /etc/apk/keys and I'm getting: `WARNING: Ignoring /home/build/packages/main: BAD signature` 2022-04-21 18:18:20 Are there replacements to systemd-hwdb? I need to set a rotation matrix for iio-sensor-proxy, and it picks it via udev property 2022-04-21 18:18:35 Or do I need to write a udev rule? 2022-04-21 18:23:56 if that's the hardware id database then you're probably looking for the hwids package 2022-04-21 18:25:40 aha figured it out, it's cause I ran `abuild-sign` which added the signature multiple times, removing the index file and running `abuild index` did it 2022-04-21 18:31:32 durrendal: thank you 2022-04-21 18:34:56 :) happy to help 2022-04-21 18:59:40 durrendal: do you by any chance have the time to bump sbcl? someone was asking about it being put into edge as well, since technically the 3.12 version is newer :p 2022-04-21 18:59:59 My terminal is borked: https://imgur.com/a/Elk479E 2022-04-21 19:00:41 Alpine 3.15.4 standard, booted from usb flash, it should be intel graphics 2022-04-21 19:02:21 is that the braille upgrade? 2022-04-21 19:02:21 :p 2022-04-21 19:03:35 Hehe 2022-04-21 19:03:43 matrix mode 2022-04-21 19:07:32 ikke: matrix mode in which deja vus appear 2022-04-21 19:19:35 nomodeset didn't help 2022-04-21 19:20:43 it works not on alpine? 2022-04-21 19:21:05 uh... it works on not alpine? 2022-04-21 19:21:26 with arch linux and ubuntu, graphics work 2022-04-21 19:23:19 but i didn't try 5.15 kernel on these systems, though 2022-04-21 19:25:41 ok. it kind of looks like bad cable but if it works on other linux then probably not 2022-04-21 19:29:53 Gonna try 3.14 2022-04-21 19:30:08 I suspect driver brokerage 2022-04-21 19:33:37 Yeah, 3.14 works 2022-04-21 19:35:53 odd that nomodeset wouldn't work 2022-04-21 19:36:22 that seems too high resolution to be caused by vbios 2022-04-21 19:36:34 is it bios or efi? 2022-04-21 19:36:41 Efi 2022-04-21 19:36:55 Secure boot is off fwiw 2022-04-21 19:37:31 psykose: I appreciate you trying to bump it recently, I started looking into the issue when I saw that it failed on the builders 2022-04-21 19:37:54 but I've bad pretty miserable luck getting it to build 2022-04-21 19:38:35 yep 2022-04-21 19:38:45 and i can't test it locally since for some reason it spawns an interactive shell :p 2022-04-21 19:38:52 and i forgot how to get it to not do that 2022-04-21 19:38:56 Also root filesystem isn't found, and I'm in initramfs shell 2022-04-21 19:39:16 Eric Timmons has some notes on the build issue which seem to indicate that it's a musl libc incompatibility issue that occurred between whatever version is in 3.15 and edge 2022-04-21 19:39:17 take your time, though! was just wondering if you knew about it 2022-04-21 19:39:42 if you mean musl directly, it's the last 1.2.3 musl release with some hundreds of commits 2022-04-21 19:40:31 Ermine: maybe try video=efifb:off or csm boot 2022-04-21 19:40:40 the project that has been absolutely destroying my free time is coming to an end (thank god) so I'm trying to get back in the flow of things. Thank you so much for all of the maintenance you've done on my packages, it is sincerely appreciated :) 2022-04-21 19:40:53 hahaha hundreds of commits, I have my work cut out for me 2022-04-21 19:42:36 well, anything over some-2-digit-amount is always the same ~8 bisect steps :p 2022-04-21 19:42:39 or whatever the numbers wer 2022-04-21 19:42:40 were* 2022-04-21 19:42:43 i'm sure it'll be okay 2022-04-21 19:43:19 that's true, I'll make sure to mix a stiff drink before I get started though 2022-04-21 19:43:53 :3 2022-04-21 19:49:32 Hello71: nope, this didn't work 2022-04-21 19:51:21 And my device doesn't have csm 2022-04-21 19:51:40 Ok, gonna stick with 3.14 for now 2022-04-21 19:52:51 Anyway apparently I need to iron out kernel bugs on this device 2022-04-21 20:58:57 upgrading to linux-lts causes my system to no longer boot and I think ive tracked the issue down to a kernel config change 2022-04-21 21:00:21 was it in the past 5-7 days 2022-04-21 21:00:29 and were you booting efistub :p 2022-04-21 21:00:43 yup 2022-04-21 21:00:45 #13728 2022-04-21 21:00:57 yeah, thats what I noticed 2022-04-21 21:01:01 seemingly there was an issue with the new config system, and so we.. don't have it set in a recent upgrade 2022-04-21 21:01:13 is that intended 2022-04-21 21:01:21 or willl efi_stub get reenabled again 2022-04-21 21:01:31 it's not intended at all, no 2022-04-21 21:01:41 I use UEFI to boot directly instead of a bootloader so yeah 2022-04-21 21:02:40 well, its good to know that it is known. I will just have to wait 2022-04-21 21:03:14 broken_boot: living life on the Edge? ;-) 2022-04-21 21:03:25 Yes 2022-04-21 21:03:29 dont remember why though 2022-04-21 21:03:35 (Think I need some firmware files maybe?) 2022-04-21 21:03:56 as an aside, I seem to have lost an initramfs setting while up up/downgrading my kernel 2022-04-21 21:04:31 do you know where id specify a driver to be loaded early when using mkinitfs 2022-04-21 21:04:59 ext4 driver isn't loading after the nvme driver for what ever reason so I have to modprobe it manually in rescue shell 2022-04-21 21:45:08 broken_boot: where on boot your system hangs ? Reading you and because my system hangs on boot am trying to diagnose during the last week and not findings the problem, wondering if I am not on the same issue. (Would be strange as I am using grub) 2022-04-22 01:00:00 blt[m] (broken_boot too if still here with another nick): upgrading to 3.15 broke initrd when updating without -a, rerun apk update && apk upgrade -a to fix 2022-04-22 01:01:07 (hm probably different for broken_boot, didn't read far back enough. well, he's not here anymore anyway) 2022-04-22 02:37:41 Is efistub supposed to work with a compressed kernel? When I was testing it previously I had to make sure to uncompress the kernel to make it work. 2022-04-22 02:38:09 yes 2022-04-22 02:38:45 if you mean the literal kernel image, that comes from the repos, that is already compressed, then yes (even though it's currently broken in edge) 2022-04-22 02:38:48 there is no uncompressing to do 2022-04-22 02:38:57 if you mean you manually ran gzip on the kernel.. well 2022-04-22 02:39:34 Hmm, weird, when I installed the kernel from the repos, then booted into the EFI shell, I couldn't execute the kernel image. 2022-04-22 02:39:48 the current one right now is broken and not an efistub 2022-04-22 02:40:03 before (a week ago) it should work, so i assume it would be another efi issue 2022-04-22 02:40:24 It worked after uncompressing, though 2022-04-22 02:41:17 Do you think it could be due to it being QEMU UEFI, or is booting compressed loaders a standard in UEFI? 2022-04-22 02:41:54 how did you uncompress it? 2022-04-22 02:42:43 gunzip I think 2022-04-22 02:43:01 https://img.ayaya.dev/PIc3QruCsIpH.png 2022-04-22 02:43:07 did you compress it by hand first..? 2022-04-22 02:43:33 No 2022-04-22 02:44:44 i'm pretty sure the compression is on specific parts of the data after the efi stub, so you can't just run gunzip 2022-04-22 02:45:27 there's tools like extract-vmlinux for that, but if it's not those i'm not really sure what you did 2022-04-22 02:45:55 and the kernel decompresses itself at load, i don't think it's relying on anything in efi 2022-04-22 02:52:25 I used the edge kernel (but a few weeks ago, 5.16.14-r1), and running file on it says it's gzip compressed data 2022-04-22 02:53:58 uhh 2022-04-22 02:54:01 could you show that 2022-04-22 02:56:19 https://0x0.st/oTA2.png (the .bak file is the original from apk, I gunzip-ed it to be bootable) 2022-04-22 02:59:40 it's.. not the original file then? 2022-04-22 02:59:52 shrug 2022-04-22 03:00:30 if i rename the files or something and diff them then nothing happens, but i bet you will get an actual difference 2022-04-22 03:01:29 I don't understand what you mean 2022-04-22 03:01:38 if you diff them are they the same :p 2022-04-22 03:02:03 and what do you mean by 'original from apk'? 2022-04-22 03:03:06 The vmlinuz-edge.bak file was renamed from vmlinuz-edge after installing linux-edge with apk. Then I gunzip-ed vmlinux-edge. When I diff them they _do_ differ. 2022-04-22 03:06:31 maybe the aarch64 kernels work completely differently, no idea 2022-04-22 03:07:27 ah, seems they are indeed just a gzip 2022-04-22 03:07:28 hm 2022-04-22 03:10:14 ah, right 2022-04-22 03:10:16 Weird, since it's the x86_64 config that sets CONFIG_KERNEL_GZIP=y, whereas those lines isn't even in the aarch64 config. 2022-04-22 03:10:19 on aarch64 they can't self-decompress 2022-04-22 03:10:32 the booting thing has to do it for you 2022-04-22 03:10:40 so if it doesn't, you do indeed need to do it yourself 2022-04-22 03:11:45 i.e. things like u-boot support that, but yeah, as a bare efistub i think it won't work at all, since the compression is just literally a gzip added-on after as you see 2022-04-22 03:11:50 sorry, i really just assumed x86 :p 2022-04-22 03:11:56 Ah, I see. Not currently using EFISTUB booting, but good to have understood. There's some way to run hooks after the kernel is installed/upgraded? 2022-04-22 03:12:37 etc/apk/commit_hooks.d can have executable scripts that get run after every operation 2022-04-22 03:12:56 that take an argument of whether it's pre-commit or post-commit 2022-04-22 03:12:59 in that sense, sure 2022-04-22 03:13:56 Thanks 2022-04-22 07:01:09 Kernel does not react to echo 1 > /sys/class/graphics/fbcon/rotate 2022-04-22 07:01:35 This is alpine 3.14.6 while being in initramfs shell 2022-04-22 07:09:38 Asmadeus: thanks! let me find my old post for you as I dont believe to have the same issue 2022-04-22 07:12:10 Asmadeus: Hi, trying to get some help as I am stuck since last week on a RAID issue. I set up a cloud based OVH bare metal using IPMI and chroot to install Alpine instead of proposed Linux version. Everything is working ok but when I reboot I am stuck on boot process. I am able to get the error : /dev/md0 on /sysroot failed : No such file or directory. I checked my mdadm, my modules and added raid , did a mkinitfs, fstab is ok, UUID as 2022-04-22 07:12:10 well. Just before this error i got a md/raid1:md0 : active so it seems raid module is loaded and md0 active. I also tried to add to openrc boot mdadm and mdadm-raid but no more success. Any clue? 2022-04-22 08:34:12 Just the name that's not the same? iirc you should get an emergency shell, can you access it from the ipmi ? 2022-04-22 08:34:20 blt[m]: ^ 2022-04-22 12:35:58 which kernel version will be packed with 3.16? 2022-04-22 12:36:30 The one that is in edge now 2022-04-22 12:36:43 5.15.x 2022-04-22 12:37:17 I haven't heard about a newer LTS kernel 2022-04-22 12:40:12 Ok, thank you 2022-04-22 12:44:23 Hello. I can't make xorg work on virt-manager. Unable connect to X server connection refused. Server terminated with error (1). Xorg log says xf86EnableIO failed enable I/O (Operation not permited). ACPI failed (no such file or directory). I installed xf86-video-modesetting and xf86-video-qxl and edited /etc/X11/xorg.cong eccording to qemu guide. xf86-video-sis and 2022-04-22 12:44:25 xf86-input-synaptics. apk complaines about gdk-pixbuf not available so I downloaded and installed it manually. 2022-04-22 12:46:41 how do you start X? via startx or display manager? 2022-04-22 12:47:21 startx 2022-04-22 12:47:56 I use startx. 2022-04-22 12:49:09 .xinitrc with xterm 2022-04-22 12:52:27 you can try to copy .xinitrc to root and execute startx as root. Not sure if this is a good solution 2022-04-22 12:53:40 Another option is to install elogind, start it with rc-service elogind start, re-login and start x 2022-04-22 12:54:12 u0: is you user member of video and input? 2022-04-22 12:54:13 Does it metter what build to use if I want xorg? Extended or virtual. Or they both shoud work the same? 2022-04-22 12:54:28 Yes user is a memeber of thouse groups. 2022-04-22 12:54:32 u0: the only difference is what is present on the iso 2022-04-22 12:54:46 so you use extended if you need more to be able to install alpine 2022-04-22 12:54:52 but the end result should not be different 2022-04-22 12:55:25 What alpine package contains the 'w' command? 2022-04-22 12:55:38 https://pkgs.alpinelinux.org/contents?file=w&path=&name=&branch=edge 2022-04-22 12:56:25 Thanks. Assumed that searching for a single letter would just return everything containing a 'w'! My bad, thanks ikke 2022-04-22 12:56:39 No, it's strict by default 2022-04-22 12:56:49 apk does that 2022-04-22 12:57:48 apk search -x cmd:w 2022-04-22 13:01:28 Ah, but the 'w' command doesn't seem to work. 2022-04-22 13:01:46 something something utmps 2022-04-22 13:01:48 https://paste.debian.net/1238675/ 2022-04-22 13:01:59 fix is on the way afaik 2022-04-22 13:02:53 adhawkins: try installing utmps and then relogin 2022-04-22 13:04:06 Still the same ikke 2022-04-22 13:04:23 it doesn't work currently 2022-04-22 13:04:29 can confirm 2022-04-22 13:04:53 Ah ok, fair enough! Cheers panekj and Ermine 2022-04-22 13:07:55 Hmmm, this is odd. 2022-04-22 13:07:59 Got this line in my fstab: 2022-04-22 13:08:14 nas.gently.org.uk:/mnt/data/music /backups/music nfs ro 0 0 2022-04-22 13:08:34 However, if I umount it, and then mount it again (umount /backups/music; mount /backups/music) it's still showing as rw 2022-04-22 13:08:48 nas.gently.org.uk:/mnt/data/music on /backups/music type nfs4 (rw,relatime,vers=4.2,rsize=524288,wsize=524288,namlen=255,hard,proto=tcp,timeo=600,retrans=2,sec=sys,clientaddr=192.168.202.168,local_lock=none,addr=192.168.202.104) 2022-04-22 13:11:45 I've also tried 'mount -o remount,ro /backups/music' and that doesn't seem to work either. 2022-04-22 13:11:58 Same command works on Debian for the same mount point. 2022-04-22 13:12:50 do kernel logs say anything about this mount? 2022-04-22 13:13:39 Nothing in dmesg since the system booted. Is that what you were referring to? 2022-04-22 13:13:52 Last entry is 16.9 seconds after boot (which is ages ago) 2022-04-22 13:14:07 what bad signature means in apk output? 2022-04-22 13:14:08 Okay 2022-04-22 13:14:39 panekj: it means the signature in the index does not match the signature of the package 2022-04-22 13:14:51 panekj: what package is giving that? 2022-04-22 13:15:02 my own 2022-04-22 13:15:11 but abuild didn't complain 2022-04-22 13:15:14 It's currently doing a backup from that mount Ermine, so when that's complete I'll try a reboot. 2022-04-22 13:18:12 panekj: did you add your public key to /etc/apk/keys? 2022-04-22 13:18:39 yes 2022-04-22 13:19:17 did you rename your public key? It should match the private key name as it was when you built the package 2022-04-22 13:20:31 it's same name 2022-04-22 13:21:06 What command exactly gives that error? 2022-04-22 13:21:46 apk fix, reinstalled package due to APKBUILD update 2022-04-22 13:23:08 and what is the error? 2022-04-22 13:24:17 ERROR: systemd-udevd-250-r0: BAD signature 2022-04-22 13:25:33 panekj: what if you run abuild index in the package directory? 2022-04-22 13:26:54 still same 2022-04-22 13:29:06 ok, seems like stale cache, works with --no-cache 2022-04-22 13:30:21 ok, so you didn't run apk update 2022-04-22 14:18:04 Hmmm...after a reboot, the mounts are still showing as rw. 2022-04-22 14:18:58 dus nfs4 support ro? 2022-04-22 14:19:03 https://paste.debian.net/1238683/ 2022-04-22 14:19:26 ikke: If I do the same on Debian I can remount it ro 2022-04-22 14:19:37 adhawkins: hmm, ok 2022-04-22 14:20:58 This is a similar attempt on Debian: https://paste.debian.net/1238684/ 2022-04-22 14:50:08 Seems to be an issue with busybox mount. After installing util-linux-misc it works as expected. 2022-04-22 14:50:39 aha, ofcourse 2022-04-23 03:20:21 Trying to package lightdm-gtk-greeter-settings from Xubuntu, and get this when running 2022-04-23 03:20:31 ACTION sent a code block: https://matrix.org/_matrix/media/r0/download/matrix.org/AvrzcURDLPYbNGYJsfsczRhc 2022-04-23 03:20:37 I assume this is non-trivial to fix? 2022-04-23 04:02:07 you can replace it with py3-gettext and replace the import with import gettext or whatever 2022-04-23 04:20:30 Oooh, really? That opens some doors 2022-04-23 08:44:13 Hi. How do I switch keyboard layout while running alpine-setup. I chose 'th' but need to switch back to 'en' in order to set the timezone. Tried Alt+Shift, Alt+Space, Alt+CapsLk, but didn't work. Thanks. 2022-04-23 09:29:37 you can run setup-keymap and setup-timezone later 2022-04-23 10:22:33 what could I use instead of xdm? xdm starts but I cant type. input works in xinit however ... 2022-04-23 10:22:48 genericum: did you add your user to the input group? 2022-04-23 10:23:08 my user is fresh 2022-04-23 10:23:13 so no 2022-04-23 10:24:02 add it to input, video and audio 2022-04-23 10:24:13 and make sure you login again 2022-04-23 10:24:18 but xdm is startet by root usually 2022-04-23 10:29:21 added root to the input group but still I cant type 2022-04-23 10:29:26 mouse is working 2022-04-23 10:29:29 as before 2022-04-23 10:29:57 @ikke 2022-04-23 10:30:54 for root it would not matter 2022-04-23 10:31:00 it would have access anyway 2022-04-23 10:31:11 How do you start xdm / is xdm started? 2022-04-23 10:31:58 xdm service I gather? 2022-04-23 10:33:27 is xf86-input-libinput installed? 2022-04-23 10:35:13 i guess so ... as I said if i start x using startx there is no input problem 2022-04-23 10:36:01 startup via /etc/init.d/xdm start 2022-04-23 10:37:49 maybe verify 2022-04-23 10:37:55 apk version xf86-input-libinput 2022-04-23 10:40:29 yes 1.2.0-r1 2022-04-23 10:40:38 ok 2022-04-23 10:40:55 and just using startx is not an option? 2022-04-23 10:41:05 i would always be user 2022-04-23 10:41:11 root 2022-04-23 10:41:34 You'd login as you user and then run startx 2022-04-23 10:41:43 as your* 2022-04-23 10:41:45 /etc/X11/xdm/xdm.conf doesnt give me any clues 2022-04-23 10:42:03 I haven't used xdm, so not really sure what is causing it 2022-04-23 10:42:10 what could I use instead ? 2022-04-23 10:42:25 lightdm doesnt start somehow 2022-04-23 10:42:45 I just use startx directly 2022-04-23 10:44:12 that kinda sucks I'd have to adjust all the permissions for tty and card0 2022-04-23 10:44:25 Not if you are in video 2022-04-23 10:44:43 I do that exactly myself 2022-04-23 10:44:49 didn't need to adjust any permissions 2022-04-23 10:46:13 How does xdm work? Does it use a running X server? If so, /what do the logs say/? 2022-04-23 10:46:58 I gont the error 2022-04-23 10:47:09 i was actually typing but blue on blue 2022-04-23 10:47:15 oh ha! 2022-04-23 10:47:49 Xresources.alpinelinux was configured that way by default 2022-04-23 11:28:28 anyone know if it's possible to install over a running installation? I created a chroot in tmpfs with alpine-base installed, but setup-disk refuses to install to /dev/vda, which is where the running system is on 2022-04-23 11:30:28 Trying to install alpine on a VPS that does not offer it, but also have no option to load a custom iso 2022-04-23 11:33:38 Mount that tmpfs as / and umount everything else. Then you have free reign over the partitions. (Then you notice you lost your network connection and forgot to install essential packages--after you reamed the disk.) 2022-04-23 11:34:55 humm: :) 2022-04-23 11:35:18 How would I mount that tmpfs as /? 2022-04-23 11:37:35 humm: with switch_root? 2022-04-23 11:39:09 I don't know 2022-04-23 11:39:47 Let me reboot the vps to check if they have some kind of emergency console 2022-04-23 11:40:53 perhaps mount -o move /mnt / 2022-04-23 11:41:53 that won't umount the old / 2022-04-23 11:56:11 There is a 1G swap partition I could abuse 2022-04-23 14:21:36 ikke: I've not closely examined or used it, but there is https://github.com/ptrcnull/alpine-infect/blob/master/infect.sh 2022-04-23 14:22:42 interesting 2022-04-23 14:22:54 some gdb magic 2022-04-23 14:23:16 ptrc: How tested is ^? :P 2022-04-23 14:25:05 right not I've installed alpine on the 2nd disk 2022-04-23 14:25:07 now* 2022-04-23 14:26:22 trying to figure out how to get grub boot that 2022-04-23 14:26:46 /boot on the default OS is not a separate partition 2022-04-23 14:36:14 there are various strategies for this including re-execing init and installing a temporary initramfs 2022-04-23 14:36:48 this gdb method doesn't seem that good, it won't work with statically linked init for example 2022-04-23 14:37:11 although typically systemd would be the source so that shouldn't be a huge issue 2022-04-23 14:38:02 but if you're going to rely on systemd then why not use systemctl switch-root 2022-04-23 14:39:02 can that switch to non-systemd inits? 2022-04-23 14:39:03 in general i think the problem with this type of thing is that it's almost impossible to recover from early boot issues without the option to boot from external media 2022-04-23 14:39:26 yeah, it's going to be tricky 2022-04-23 14:39:30 I'm aware of that 2022-04-23 14:39:53 If something fails, I can reinstall the vps 2022-04-23 14:40:01 i vaguely recall trying it in the past, but can't quite recall. the man page strongly implies that it can though: "This call takes two arguments: the directory that is to become the new root directory, and the path to the new system manager binary below it to execute as PID 1. If the latter is omitted or the empty string, a systemd binary will automatically be searched for and 2022-04-23 14:40:03 used as init. If the system manager path is omitted, equal to the empty string or identical to the path to the systemd binary, the state of the initrd's system manager process is passed to the main system manager, which allows later introspection of the state of the services involved in the initrd boot phase." 2022-04-23 14:41:09 the other issue that i have with this kind of thing is that such restricted VPSs are usually so crappy that they're not worth using anyways 2022-04-23 14:42:52 there's also https://github.com/gh2o/digitalocean-debian-to-arch, https://github.com/drizzt/vps2arch 2022-04-23 14:43:31 I'm now trying to get dual boot to work 2022-04-23 14:44:51 some very ancient xen providers don't actually use the grub, the kernel is loaded from host 2022-04-23 14:45:09 well, actually some crappy kvm providers too (*cough* digitalocean *cough*) 2022-04-23 14:45:22 so you need to replace /sbin/init 2022-04-23 14:46:36 I think their previous platform worked like that, but it appears the new platform is actually using grub 2022-04-23 14:46:56 I can use grub and choose what it boots 2022-04-23 14:49:09 I already am getting into the initramfs 2022-04-23 14:55:23 fwiw that doesn't necessarily mean it's using the guest grub, it could be a host grub (e.g. pvgrub). but that doesn't sound like your issue 2022-04-23 14:56:21 I can update it from the guest and I see the changes 2022-04-23 14:57:03 right, it uses the guest config, but the actual binary is stored on the host and can't be updated 2022-04-23 14:57:18 i think this is only an issue if you want to use some exotic filesystem 2022-04-23 14:57:34 right, but that's not the case for me 2022-04-23 14:57:55 mmhmm 2022-04-23 14:58:07 ah, os-prober now found my alpine install 2022-04-23 14:58:19 let's see 2022-04-23 15:04:45 is it expected for /dev/disk to be world writeable (chmod 777)? clean install from ISO 3.15.4 2022-04-23 15:05:37 noticed this on a server I built recently with 3.15.2, so repeated with 3.15.4 in a VM to double check, and both have world-writeable /dev/disk 2022-04-23 15:05:55 /dev/disk is chmod 755 on other distros I'm running 2022-04-23 15:17:14 hmm, there is no /dev/disk in cleanly installed 3.13 and 3.14. So I guess this is an issue with 3.15? 2022-04-23 15:17:38 crono_log: By default mdev does not add these entries 2022-04-23 15:18:00 udev does, but I believe it has been added to mdev as well, though cannot find where 2022-04-23 16:28:35 https://gitlab.alpinelinux.org/alpine/aports/-/blob/master/main/busybox-initscripts/mdev.conf#L40-48 2022-04-23 16:29:13 i guess busybox is invoking it with umask 0 2022-04-23 19:41:49 Hello71: I managed to install alpine on the 2nd disk and boot into it 2022-04-23 20:16:35 Hello71: Booted into alpine on the main disk now :D 2022-04-23 20:16:43 :) 2022-04-23 20:31:42 ikke: (You might've been able to replace the running system. Compile apk, overwrite system. Clean up old system. Reboot.) 2022-04-23 20:32:44 s/\(Clean up old system\.\) \(Reboot\.\)/\2 \1/ 2022-04-23 20:34:34 humm: You can use apk-static, no need to compile it :) 2022-04-23 20:34:44 But now I could just use alpine-setup 2022-04-23 20:36:43 Has anyone gotten Alpine booting with an initramfs generated by booster yet? 2022-04-23 20:37:03 not yet 2022-04-23 20:51:55 damn ok 2022-04-23 21:29:28 it's been in testing for 4 months already, probably at least one person has tested it? 2022-04-23 22:46:13 i would imagine nmeum uses it actively 2022-04-23 22:50:42 psykose: If I'm doing the py3-gettext patching right, will this basically be whack-a-mole of replacing every import in every .py file one by one? 2022-04-23 22:51:04 sure, but you can just grep for the imports 2022-04-23 22:54:35 Ahh... instead of doing the patch manually? 2022-04-23 22:54:44 That sounds smarter, but harder with my current toolset 2022-04-23 22:55:25 i do mean manually 2022-04-23 22:55:29 but you can see where they are 2022-04-23 22:55:37 it's not just picking out random files until it works 2022-04-23 22:55:38 :) 2022-04-23 22:56:40 I feel attacked haha 2022-04-24 02:23:41 Does anyone have the proper inter-dependencies to get kvm/qemu/libvirt playing nice on an Alpine host so I can try and make the MR for adjusting the services? 2022-04-24 02:23:59 There must be some timing/race thing going on, because which services crash or work changes from boot to boot here and it is reliable maybe 5% of the time 2022-04-24 02:29:01 use qemu-openrc instead /s 2022-04-24 02:55:42 Eh? haha 2022-04-24 02:56:26 Are OpenRC service files done by the distro, or upstream? So like, would Artix OpenRC be a good place to look and see if they have tackled this alread? 2022-04-24 03:31:56 theoretically either 2022-04-24 03:32:06 in practice openrc is not very popular so mostly distro 2022-04-24 03:32:31 i suspect libvirt is not very popular on alpine 2022-04-24 03:33:01 either qemu-openrc for non-interactive use or direct qemu invocation for interactive 2022-04-24 05:10:35 Me using virt-manager is just adding more problems to the pile? 2022-04-24 05:11:17 I know Alpine is typically hosted, so I guess it makes sense it being a host is less explored, but it's an important thing for me to iron out for Alpine Linux on (my) desktop 2022 2022-04-24 09:22:39 We run alpine on all of our bare metal hosts 2022-04-24 09:24:05 But if we run vms, we use qemu, not libvirt 2022-04-24 13:38:35 Apparently I need to read up more on this because my understanding was libvirt basically uses qemu to do the heavy lifting, so I thought they were the same thing 2022-04-24 13:38:38 Hmm, back I go 2022-04-24 13:44:47 libvirt is "middleware": it can run qemu with defined configuration, provide client with spice/qemu connection infos, ... 2022-04-24 13:44:59 libvirt also can do lxc (and I think xen and more?) 2022-04-24 13:45:48 I think I've seen it have support for virtualbox, even 2022-04-24 13:54:42 ikke: how are you using qemu? are you executing them in tmux? 2022-04-24 13:54:53 qemu-openrc 2022-04-24 13:55:00 but sometimes in tmux as well :) 2022-04-24 13:56:46 thanks, im checking it out 2022-04-24 13:56:46 so far i've written my shell scripts to template the long command from a few variables, but it's not yet very reliable^^ 2022-04-24 14:08:23 https://libvirt.org/drivers.html "The hypervisor drivers currently supported by libvirt are: LXC - Linux Containers, OpenVZ, QEMU, Test - Used for testing, VirtualBox, VMware ESX, VMware Workstation/Player, Xen, Microsoft Hyper-V, Virtuozzo, Bhyve - The BSD Hypervisor, Cloud Hypervisor" 2022-04-24 14:38:18 wondering if there is an option to setup-alpine during the disk step (sys,data..) to detect that nvme are in use and so to have the script to add nvme to /etc/mkinitfs/mkinitfs.conf before launching mkinitfs ? this would save extra step at the end of the script and potentialy avoid non booting system because of nvme module not loaded.. 2022-04-24 14:40:54 https://gitlab.alpinelinux.org/alpine/alpine-conf/-/blob/master/setup-disk.in#L533 2022-04-24 14:41:30 It's already there 2022-04-24 14:41:44 ikke: indeed I see that, I am wondering why it is not working on my case 2022-04-24 14:42:06 It depends on /dev/nvme to be present 2022-04-24 14:42:08 could it be raid 1 on top of nvme drives the issue? 2022-04-24 14:42:29 Does `ls /dev/nvme*` return something? 2022-04-24 14:42:56 yep 2 drives 2022-04-24 14:43:28 nvme0n1 and nvme1n1 2022-04-24 14:45:10 I noticed that if I dont do a modprobe -r nvme & modprobe nvme , even if the drives are detected in lsblk, I think they were not listed in /dev, perhaps related 2022-04-24 14:45:14 blt[m]: https://gitlab.alpinelinux.org/alpine/alpine-conf/-/issues/10502 2022-04-24 14:45:31 and I don't see something in the code to do a modprobe nvme before 2022-04-24 14:46:34 ikke: thanks ! same problem without LVM 2022-04-24 14:47:08 will try a fresh install doing a modprobe nvme at the beggining to see 2022-04-24 15:23:00 satmd: Hello71 apparently my penchant for GUI tools is biting me. I need to re-think my workflow here 2022-04-24 15:34:30 hello 2022-04-24 15:34:57 ah 2022-04-24 17:20:00 ikke: so if I do a modprobe nvme before launching the setup-disk it does add the nvme module ! interesting, now if I try a setup-disk -e it seems that I am missing a module (I guess cryptsetup) as I am dropped to rescue shell, I am rebooting on live cd and chrooting to investigate, but I suppose that the module should be added automatically https://gitlab.alpinelinux.org/alpine/alpine-conf/-/blob/master/setup-disk.in#L594 2022-04-24 17:23:12 hello, if I want to make the NoIR camera v2 work with a Raspberry Pi 3, is Alpine Linux able to handle it? I just need firmware blobs for that, right? 2022-04-24 17:29:15 billiob: what bootload is being used? 2022-04-24 17:29:19 blt[m]: * 2022-04-24 17:29:53 ikke: grub 2022-04-24 17:30:09 blt[m]: ok, and what firmware, bios or uefi? 2022-04-24 17:30:48 I saw it launching initram and dropped to rescue, I am manually assembling md , cryptsetup and mounting to chroot to analyze it from iso 2022-04-24 17:30:49 uefi 2022-04-24 17:31:05 Most apk commands on my system (even "apk info") do nothing for around 15 seconds and then start the actual execution. I'm running postmarketOS v21.12.4 aarch64 on Pine64 PinePhone (1.2). I've already asked in the official postmarketOS chat room but the people there had no idea what the reason for this issue could be. 2022-04-24 17:31:41 Noclip[m]1: maybe run it with strace? 2022-04-24 17:41:31 ikke: just checked and cryptsetup is in mkinitfs.conf... 2022-04-24 17:42:35 arf but no nvme :( 2022-04-24 17:45:30 ikke: In those first 15 seconds I get tons of "mmap(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0xffff887d1000". 2022-04-24 17:45:30 In between there are "read" and "brk" calls. 2022-04-24 17:48:35 Mhh, I might have found somethimg interesting! 2022-04-24 17:49:03 I first noticed those long execution times directly after installing the full ansible package. 2022-04-24 17:50:30 And now I noticed that most or all of those "read" syscalls seem to read some file which is part of the ansible community collection. 2022-04-24 17:50:56 "read(6, "lib/python3.9/site-packages/ansi"..., 131066) = 131066" 2022-04-24 17:51:53 "read(6, "_collections/community/azure/tes"..., 131031) = 131031" 2022-04-24 17:55:40 that's the file *contents* 2022-04-24 17:55:46 the file is 6 2022-04-24 17:56:18 Need to find out what opened fd 6 2022-04-24 17:58:22 -y 2022-04-24 17:58:52 but i reckon this is just /lib/apk/db/installed 2022-04-24 17:59:02 yeah 2022-04-24 17:59:30 Noclip[m]1: anything in dmesg? 2022-04-24 17:59:56 https://pkgs.alpinelinux.org/contents?page=930&arch=x86_64&branch=edge&name=ansible 2022-04-24 18:00:18 i reckon there's your problem 2022-04-24 18:00:29 It's "read(6," for all of the reads! 2022-04-24 18:01:49 Hello71: "i reckon there's your problem" 2022-04-24 18:01:49 What's that? 2022-04-24 18:02:38 And why is this casing such a long delay? 2022-04-24 18:04:19 some issue with storage?/ 2022-04-24 18:05:25 ikke: The OS is installed on the PinePhone's internal emmc storage. 2022-04-24 18:05:37 either your storage is very slow and the installed database exceeds available ram, or your cpu is very slow 2022-04-24 18:05:39 or both 2022-04-24 18:06:39 ls -lh /lib/apk/db/installed 2022-04-24 18:06:58 maybe remove ansible and check if it speeds up 2022-04-24 18:07:02 Hello71: Well the PinePhone is not a very powerful device. 2022-04-24 18:07:38 "-rw-r--r-- 1 root root 6.2M Apr 24 19:35 /lib/apk/db/installed" 2022-04-24 18:08:44 ikke: The delay first occured directly after installing ansible. 2022-04-24 18:08:45 6.2M seems quite large 2022-04-24 18:09:53 Noclip[m]1: like I suggested earlier, can you try to remove ansible to verify? 2022-04-24 18:10:49 ACTION sent a code block: https://matrix.org/_matrix/media/r0/download/matrix.org/sXpCSWjhORBrCXohIlyzyVTp 2022-04-24 18:11:24 oh wow, TIL :D 2022-04-24 18:11:30 didn't know that command existed 2022-04-24 18:12:15 ikke: Didn't know that either, I just tried `apk status` and `apk stats`. xD 2022-04-24 18:12:45 files: 81445 2022-04-24 18:14:23 I don't see GRUB_ENABLE_CRYPTODISK=y nor cryptdevice= in /etc/default/grub but I can see cryptroot and cryptdm don't know if it could be the reason of no booting (appart from missing nvme but I manualy added it and rebuild mkinitfs) 2022-04-24 18:16:09 seems like alpine:edge + ansible = 4590233 /lib/apk/db/installed 2022-04-24 18:16:45 "ansible-4.8.0-r0 installed size: 478 MiB" 2022-04-24 18:17:03 Hello71: I'm not on edge, I'm on stable. 2022-04-24 18:17:21 i don't think there's really a way around this except "don't install ansible" 2022-04-24 18:17:57 perhaps ansible could use some APKBUILD tidy up and subpackaging? 2022-04-24 18:18:20 Hello71: Why is ansible causing such a huge delay? 2022-04-24 18:18:50 lots of files on not ideal hardware 2022-04-24 18:18:51 it's tons of files which clutter lib/apk/db/installed 2022-04-24 18:20:19 also possibly apk is not as optimized when dealing with very large installed database as it could be 2022-04-24 18:20:33 People in the postmarketOS room who have way more packages with a bigger total size (disk usage of all installed packages together) installed don't have this issue. 2022-04-24 18:20:50 are they using same hardware though? 2022-04-24 18:20:52 other package managers probably suffer less from this issue because they store file lists separately 2022-04-24 18:21:09 So it's the amount of files which is causing the issues? Okay. 2022-04-24 18:21:10 and probably apkv3 will handle this better as well 2022-04-24 18:21:17 the total size doesn't matter, it's the size of the installed database which apk needs to read to know what is installed 2022-04-24 18:21:44 ikke: does it store it differently? 2022-04-24 18:21:47 panekj: "are they using same hardware though?" 2022-04-24 18:21:47 Some of them, yea. 2022-04-24 18:22:12 ikke: When will apkv3 come out? 2022-04-24 18:22:22 also theoretically we could cut the number of files in half by implementing https://gitlab.alpinelinux.org/alpine/aports/-/issues/11906#note_192289 2022-04-24 18:22:39 Noclip[m]1: Not sure, they were talking for alpine 3.16, but it will probably be the next release 2022-04-24 18:23:18 panekj: from what I understood, it's a binary format that can be directly mmapped, or something like that 2022-04-24 18:23:37 ah, something like CBOR/protobuf probably 2022-04-24 18:24:13 yeah, I was about to say if it would be better to have binary db but the downside is you can't read it with simple `cat ... | less` 2022-04-24 18:24:23 I could switch from ansible to ansible-core (which is much much smaller) and then only install those ansible modules which I really need. 2022-04-24 18:25:07 "I don't see GRUB_ENABLE_CRYPTODI..." <- auto reply, everything is ok , only raid missing in mkinit.. so the script is doing a good job but only two small thing (nvme and raid) missing to make it perfect. strange that it seems to be in the code but something is preventing to work correctly 2022-04-24 18:25:18 panekj: yes, apkv3 will be less inspectable with plain-text tools 2022-04-24 18:25:24 Cause the ansible community package is kinda like: "Here you have everything which you could possibly ever need ..." 2022-04-24 18:27:29 Thank you all for helping me troubleshooting this issue! 2022-04-24 18:48:36 is there a way to pass some argument to setup-alpine ? in my case I have raid+encryption and I don't see an alternative of doing first setup-disk -e and then setup-alpine, would it be possible to have only setup-alpine and have encryption/raid option ? 2022-04-24 18:50:27 DISKOPTS="-e" setup-alpine 2022-04-24 18:51:53 😲 I saw that but I thought it was only passed to setup-disk but now I think it is calling multiple scripts it makes sense πŸ™‚ thanks ikke ! I will test end-to-end install now 2022-04-24 18:53:18 DISKOPTS _is_ only passed to setup-disk.. 2022-04-24 18:55:10 ikke: yes as you put it on env before issuing setup-alpine I assume it will "consume" it if setup-alpine is calling setup-disk 2022-04-24 18:55:50 blt[m]: https://gitlab.alpinelinux.org/alpine/alpine-conf/-/blob/master/setup-alpine.in#L226 2022-04-24 18:56:34 It's directly passed to setup-disk by setup-alpine 2022-04-24 18:57:04 very good 2022-04-24 20:01:08 hi, anyone know if alpine is working on the rpi zero 2 w? 2022-04-24 22:33:43 ikke: I just tested and it is working great ! Do you have an idea why I still need to add manually nvme and raid to the features ? It seems that the code is supposed to handle this but something is preventing to happen 2022-04-24 22:41:19 Uefi 2022-04-25 04:40:18 blt[m]: probably due the combination of features that you are using, preventing the installer to properly detect that nvme is required 2022-04-25 04:41:50 we should really just default enable more of those 2022-04-25 04:41:56 people can always remove them after anyway if they care 2022-04-25 06:33:58 I agree 2022-04-25 08:41:43 why do i have some php7 packages installed? 'apk -s del ' doesn't show them asany deps 2022-04-25 08:41:50 …as deps of anything 2022-04-25 08:49:37 never mind, they're in world 2022-04-25 15:52:19 lopid: That raises an interesting question... Is it possible to "trim" world so that nothing gets removed except entries that are manually added (say by apk add something that was already installed by another package as a dependency) 2022-04-25 15:53:04 Saijin_Naib: world _is_ what you added manually 2022-04-25 15:53:45 Every time you run apk add, the package is added to world 2022-04-25 15:54:10 Then the resolver makes sure it is installed and all required dependencies 2022-04-25 15:54:44 So there is no analog to how apt/synaptic have a "manual" status for installed packages? 2022-04-25 15:55:23 For instance, say exiv2 is already on my machine from something else, and I forget that, then I apk add evix2. It now has a line in world and doesn't get removed even if I remove the package that originally added it as a depends 2022-04-25 15:55:53 It isn't orphaned, really, but just something I explicitly added that no longer needs that explicit line in world 2022-04-25 15:56:17 yes 2022-04-25 15:57:03 Ahh, okay 2022-04-25 15:57:11 I just need to keep it clean myself then 2022-04-25 15:57:45 tl;dr: if you add a package to world, (it's called constraint), it's up to you to make sure to remove it when you don't use it 2022-04-25 15:58:45 Awesome, thanks for clarifying 2022-04-25 16:00:21 Thankfully it is fairly easy to keep tidy 2022-04-25 16:00:23 apk doesn't try to outsmart user like apt for example, where wrong simple deletetion/addition of package, borks whole system 2022-04-25 16:00:40 Am I right that if I apk del a constraint that is required, it will get removed from world? 2022-04-25 16:00:54 yes 2022-04-25 16:01:38 panekj: Oh, believe me, I've suffered that for years... I did like how I could tell what I manually added, but yeah... Never had a release upgrade not horribly botch itself and get held dependencies or remove swathes of things I was using 2022-04-25 16:02:30 as long as something is in world that requires that package, or that package is somewhere in dependency tree, this package will remain installed but it won't be present in world 2022-04-25 16:02:45 Beautiful. I need to do that type of cleanup now 2022-04-25 16:03:21 I forgot to tag a baseline install with a virtual package, so I'm trying to get a clean one now 2022-04-25 16:04:45 apk will even tell you that you are removing package but it will remain installed as it's required by other package 2022-04-25 16:11:30 Yep! I've seen that message, and that's what prompted this. Didn't know if there was an automated way to make apk "prune" world from my mistakes haha 2022-04-25 16:32:58 do I need to be in a group to use `su`? or does su not work with accounts that dont have a password? 2022-04-25 16:42:47 depends how you configured it. wheel is popular 2022-04-25 16:44:11 Hello71 PureTryOut: yes, I use booster on my system but I still have a few things on my list that need to be addressed before I move it to community 2022-04-25 17:04:03 j`ey: su requires you to authenticate with the user you are logging into 2022-04-25 17:04:08 there is no policy around that 2022-04-25 17:04:35 (ignoring PAM) 2022-04-25 17:04:45 wheel can be used with sudo 2022-04-25 17:04:56 yes, but they asked about su, not sudo 2022-04-25 17:05:03 i was correctingly myself 2022-04-25 17:05:09 -ly 2022-04-25 17:05:14 right 2022-04-25 17:06:09 ikke: I can't seem to su root, not sure why 2022-04-25 17:06:23 j`ey: and you are using the root password? 2022-04-25 17:06:44 I have no root password 2022-04-25 17:06:50 :-O 2022-04-25 17:06:53 Then you cannot login as root with su 2022-04-25 17:07:20 lame 2022-04-25 17:07:51 You'd need to use sudo, which has heen configured to allow you to use it 2022-04-25 17:08:07 (or doas) 2022-04-25 17:37:45 not lame, sane; otherwise any Joe Hacker could do anything he likes on your system 2022-04-25 17:38:29 if it is your own system, then boot off some LiveISO, chroot and passwd 2022-04-25 17:44:28 for my use case, it's lame :P 2022-04-25 18:08:22 just log in as root then 2022-04-25 18:44:41 you can set doas/sudo to allow that specific command for whatever use case you need 2022-04-25 19:32:18 Hello everyone, I'm currently analyzing a virtual machine running Alpine Linux 3.3 2022-04-25 19:32:49 quite dated :P 2022-04-25 19:33:06 lol, yes, it's for researching proposes 2022-04-25 19:33:29 I understand there's a vulnerability (CVE-2019-5021) and if I'm not mistaken, it could be exploited before login? 2022-04-25 19:33:55 I was wondering if anyone here may be able to drop some light in this matter 2022-04-25 19:34:56 It requires the shadow package to be installed 2022-04-25 19:38:06 It's also limited to docker 2022-04-25 19:38:59 during the boot process I can see it's starting up busybox mdev, docker, hostname, keymap, networking, busybox klogd, busybox syslog, busybox acpid and busybox cron 2022-04-25 19:40:01 @ikke I can see those services starting up, but it doesn't show anything related to the shadow package 2022-04-25 19:40:15 the shadow package is not a service 2022-04-25 19:40:19 it's a package that you can install 2022-04-25 19:40:35 apk version shadow 2022-04-25 19:40:42 It probably worth a try... 2022-04-25 19:41:08 charles: first check the output of `grep root /etc/shadow` 2022-04-25 19:41:34 yes, in this case, the machine is totally locked down in the login screen, and I don't have any users, therefore I can't run any commands 2022-04-25 19:42:24 You could replace init with /bin/sh or something like that 2022-04-25 19:42:25 this is specially desing this way in order to try if the system is vulnerable to CVE-2019-5021 2022-04-25 19:43:37 Hello, I'm using the synapse package from the alpine repos to create a matrix server, but it won't start, I checked the logs and it says [Errno 99] Address not available., I then did netstat -lntu to see if there were any ports already listening on 8008 and I found none, I still tried and changed the config to start in a different port and no luck, did anyone manage to get 2022-04-25 19:43:39 synapse running using the alpine package? 2022-04-25 19:45:04 do you have a lo interface (that's up)? 2022-04-25 19:45:15 yes 2022-04-25 19:45:46 oh sorry, lol that was for @bosco 2022-04-25 19:46:09 indeed 2022-04-25 19:47:08 yes 2022-04-25 19:47:44 1: lo: mtu 65536 qdisc noqueue state UNKNOWN qlen 1000 2022-04-25 19:47:46 link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00 2022-04-25 19:47:48 inet 127.0.0.1/8 scope host lo 2022-04-25 19:47:50 valid_lft forever preferred_lft forever 2022-04-25 19:49:02 It also binds to ::1 by default 2022-04-25 19:50:33 installing it in docker and executing `python3 -m synapse.app.homeserver --config-path=/etc/synapse/homeserver.yaml` works for me 2022-04-25 19:51:17 bosco: do you have ipv6 disabled? 2022-04-25 19:52:25 On the synapse config? I couldn't find anything regarding ipv6 on the homeserver.yaml file nor the docs 2022-04-25 19:54:04 grep -A4 -- '- port' /etc/synapse/homeserver.yaml 2022-04-25 19:54:28 bind_addresses: ['::1', '127.0.0.1'] 2022-04-25 19:54:52 ahh 2022-04-25 19:54:54 right 2022-04-25 19:55:10 so I should edit the first value? 2022-04-25 19:55:50 remove that, if you don't have IPv6 2022-04-25 20:00:32 Ok, that worked 2022-04-25 20:00:34 Thanks! 2022-04-25 20:29:23 or better, don't set disable_ipv6 because it breaks the networking interface 2022-04-26 04:38:41 nmeum: cool! I have yet to manage to get it boot. For some reason it seems to produce a different type of image than the usual mkimage does, not sure what's up 2022-04-26 04:43:22 mkimage makes an iso, booster only makes initramfs 2022-04-26 04:43:45 which is one small (sort of) part of iso 2022-04-26 04:44:34 actually i doubt booster can (currently) do live usb initramfs, it probably only supports "sys" install 2022-04-26 04:45:06 oh sorry I'm confusing the things. I meant mkinitfs obviously 2022-04-26 09:53:05 with no other php packages installed, should installing php8 create a link from /usr/bin/php to /usr/bin/php8? 2022-04-26 11:50:12 PureTryOut: it should just generate a gzip compressed cpio archive by default. feel free to open a gitlab issue if you have more information about your issue. the booster initramfs generator also has a -debug flag which outputs more information 2022-04-26 11:50:45 I hope upstream releases 0.8.0 soon because that release will include more debugging facilities 2022-04-26 23:25:38 Before I open an issue/request, if anyone can correct my (mis)understanding, I am reading that the only way to have CIFS (mount.cifs) work without root is to use the pam-mount modules, which we do not currently have in alpine. 2022-04-27 00:52:00 other options include fstab "user" option 2022-04-27 00:52:11 and fuse 2022-04-27 00:52:20 and sudo/doas i guess 2022-04-27 00:57:33 afaik pam_mount is the standard-ish way if you want different mounts for each user 2022-04-27 01:55:51 Yeah, fstab user is not actually working. Once I'm in-session I have to sudo mount -a to get them to actually mount 2022-04-27 01:56:18 Yeah, what you said about pam_mount is what I was reading, which I think to me makes the most sense. Do we have that? 2022-04-27 01:56:52 I've migrated from Windows where I was using the Known Folder Redirection Policy to host my home directory on my NAS 2022-04-27 01:57:59 I'm looking to use a similar setup with Alpine, so I want my NAS folder mounted at login and not accessible to other users. I've made a simple script that xfce4-session-manager runs at login to make sure all the XDG User Directories are pointed where they want. The only thing lacking is that for some reason, I must sudo mount -a after login to get the NAS shares mounted, despite having them in fstab as user 2022-04-27 01:59:06 the man page has some caveats about nfs/cifs which i don't quite understand 2022-04-27 01:59:54 if you have a limited set of directories to mount, and the server is expected to the online at boot, it's probably better to just mount it at boot 2022-04-27 02:00:23 you don't have to mount stuff in $XDG_RUNTIME_DIR, you can mount it anywhere and just set the parent directory to only allow your user 2022-04-27 02:01:25 Just two shares from the same NAS. Both are being mounted into /media/samba/NAS/sharename 2022-04-27 02:02:29 //NAS/public/Household /media/samba/NAS/Household cifs _netdev,nofail,user,uid=1000,gid=1000 2022-04-27 02:02:29 ```//NAS/public/USERNAME /media/samba/NAS/USERNAME cifs _netdev,nofail,user,uid=1000,gid=1000 2022-04-27 02:02:44 //NAS/public/USERNAME /media/samba/NAS/USERNAME cifs _netdev,nofail,user,uid=1000,gid=1000 2022-04-27 02:03:44 They show up in Thunar after login, but they're not able to be accessed until I sudo mount -a. This is tripping me up, bad. It seems like I've followed all the recommendations needed yet it doesn't work how the manpages and SE posts all say it should 2022-04-27 02:04:02 Is it a user/group thing? A policy thing? Am I somehow missing a dependency for something? 2022-04-27 02:10:17 have you tried mount /media/samba/NAS/USERNAME 2022-04-27 02:16:24 Without sudo? 2022-04-27 02:16:25 I have not, no 2022-04-27 02:49:45 Unable to set current process capabilities: Operation not permitted 2022-04-27 02:49:45 error 2 (No such file or directory) opening credential file /home/USERNAME/.smb 2022-04-27 02:50:47 ACTION sent a code block: https://matrix.org/_matrix/media/r0/download/matrix.org/QEhDvaidmZRoEVStFCFrbykU 2022-04-27 03:32:28 not sure what that means 2022-04-27 03:40:43 I guess that whatever process is mounting can't read my .smb credentials file 2022-04-27 03:41:00 Whcih makes no sense to me since I thought that the mounting done at boot was done by root, which absolutely can read my home directory 2022-04-27 04:13:39 i thought you meant that's what happened when you ran it after logging in 2022-04-27 05:16:41 The unable to set current... is what happens when I try to run mount -a without sudo/doas 2022-04-27 05:16:57 I then stat'd the .smb credential files it claims it can't access to show that indeed, it should be accessible 2022-04-27 05:17:28 Even copying it to /root and adjusting fstab to load it from there failed, so I have no idea what "user" or group is doing the mounting at boot and why this won't work 2022-04-27 09:11:18 This isn't strictly an Alpine question, but...anyone recommend a utility I can run periodically from cron that will warn me if any 'real' disk partitions (i.e. excluding /tmp, /dev etc.) are over (say) 90% usage by sending an email? 2022-04-27 09:11:44 I realise I could script something relatively easily, but would be nice if there was a utility that could do it and allow me to (say) exclude certain partitions from the monitoring. 2022-04-27 09:45:39 monitoring applications do exactly that (on a bigger scale) 2022-04-27 09:45:55 examples: icinga, nagios, omd, zabbix, ... 2022-04-27 09:46:46 a package commonly named monitoring-plugins (dunno for alpine) contains a tool check_disk which you could use for the actual checking 2022-04-27 09:47:11 it'd return 0 for OK, 1 for WARNING or 2 for CRITICAL and you can give it arguments for which disks, which warning levels, ... 2022-04-27 09:48:02 or you go the "metrics" way, like collecting information with ..._exporter, grafana and lots of stuff 2022-04-27 09:48:47 a really compact thing is "netdata" but configuring its alarms is kinda more challenging - it'll also notice when a disk gets full _fast_ 2022-04-27 09:49:19 I'm new to alpine and don't know which of those ideas are possible with alpine, which packages currently exist 2022-04-27 09:49:40 check_disk allows for excludes for your second part of the question 2022-04-27 13:08:16 satmd: Yeah, didn't want to go the whole hog and install a full-blown monitoring system. 2022-04-27 13:10:31 check_disk looks interesting satmd, thanks for that. 2022-04-28 02:07:59 Hello71: Do we have pam_mount in Alpine? Is that something I can open an Issue/Feature Request for? 2022-04-28 02:08:37 probably not; i guess you can 2022-04-28 02:37:56 Does that fit Alpine goals and wants, or that will come up in the Issue? We have other PAM stuff already, so it seems like a good fit to me... 2022-04-28 03:37:48 i don't know of any huge issues with it 2022-04-28 04:14:35 Thanks for the guidance 2022-04-28 09:27:23 Hi everyone! 2022-04-28 09:28:48 I'm doing some penetration testing on an Alpine Linux... and I'm trying to create a root privileged user using this 1 line command: 2022-04-28 09:28:52 echo charles:$((mkpasswd -m SHA-512 password) 2>/dev/null):0:0::/:/bin/bash >> /etc/passwd 2022-04-28 09:29:12 but it seems it doesn't like my bash... 2022-04-28 09:29:53 why are there two (()) 2022-04-28 09:30:03 missing $ near second group 2022-04-28 09:30:11 I had some reading and I saw that it actually uses busybox and ash... 2022-04-28 09:30:14 don't need it at all 2022-04-28 09:30:58 but yes, to get bash you need to `apk add bash` 2022-04-28 09:31:00 charles: do you have bash installed and did you enter it manually or replaced it? 2022-04-28 09:31:21 I don't have access into the box, aka I can't login. 2022-04-28 09:31:41 chances are, it's not running bash 2022-04-28 09:31:58 I'm executing the commands thru a vulnerability I found on a Jenkins which is running 2022-04-28 09:32:31 is there any way to perform the same action using ash? 2022-04-28 09:32:57 it will probably work fine if you drop second pair of () and use /bin/sh instead of /bin/bash 2022-04-28 09:33:53 mmm 2022-04-28 09:33:56 something like this? 2022-04-28 09:34:17 echo charles:$(mkpasswd -m SHA-512 password) 2>/dev/null:0:0::/:/bin/sh >> /etc/passwd 2022-04-28 09:34:35 $(mkpasswd -m SHA-512 password 2>/dev/null) 2022-04-28 09:35:30 $(mkpasswd -m SHA-512 password 2>/dev/null):0:0::/:/bin/sh >> /etc/passwd ? 2022-04-28 09:35:57 with `echo charles:` before that, yes 2022-04-28 09:36:04 so 2022-04-28 09:36:37 echo charles:$(mkpasswd -m SHA-512 password 2>/dev/null):0:0::/:/bin/sh >> /etc/passwd 2022-04-28 09:36:51 let me give it a try... :D thanks a lot, I'll report you back in 10 minutes. 2022-04-28 09:41:33 mmm... I'm having the same error message for some reason 2022-04-28 09:41:36 illegal string body character after dollar sign;<br>&nbsp;&nbsp; solution: either escape a literal dollar sign &quot;\$5&quot; or bracket the value expression " 2022-04-28 09:41:57 this comes from the vulnerable Jenkins implementation 2022-04-28 09:42:24 this is an issue in your injection and not related to the shell 2022-04-28 09:42:35 why not just execute the mkpasswd locally and then manually paste it into your exploit command? 2022-04-28 09:43:25 (and as the help tells you, you have to escape the characters) 2022-04-28 09:43:37 i.e. \$ for every dollar sign 2022-04-28 09:44:04 @psykose, yes, it's 100% related to the injection and not the command. 2022-04-28 09:44:16 tho the previous command was faulty as well... 2022-04-28 09:44:34 I'll try again with this 2022-04-28 09:44:52 echo charles:\$(mkpasswd -m SHA-512 password 2>/dev/null):0:0::/:/bin/sh >> /etc/passwd 2022-04-28 09:47:30 nop, thanks guys!!!! 2022-04-28 09:47:48 dead end, it doesn't like the injection 2022-04-28 09:47:51 "Caused by: java.lang.Exception: charles:$(mkpasswd -m SHA-512 password 2>/dev/null):0:0::/:/bin/sh >> /etc/passwd" 2022-04-28 09:48:34 Thanks psykose and panekj! 2022-04-28 09:48:40 wish you a good day :) 2022-04-28 09:50:03 :) 2022-04-28 10:00:07 I'm having an ACL issue with nextcloud 2022-04-28 10:00:40 ACTION sent a code block: https://matrix.org/_matrix/media/r0/download/matrix.org/mlfElZYSiwxCMTRuRNLrDIdN 2022-04-28 10:00:51 `doas -u nextcloud less /etc/nextcloud/config.php 2022-04-28 10:00:51 ` 2022-04-28 10:00:51 /etc/nextcloud/config.php: Permission denied 2022-04-28 10:01:06 can someone explain why 2022-04-28 10:01:25 the directory is missing +x 2022-04-28 10:01:44 chmod 770 /etc/nextcloud 2022-04-28 13:54:29 Saijin_Naib: actually, i forgot, the "enterprise" solution for this problem is to mount on boot and then use either uid (nfs sec=sys) or kerberos auth 2022-04-28 13:59:38 Hello71: That is quite the fix, haha. Probably out of scope for a small home deployment πŸ˜› But something to think about none the less 2022-04-28 18:29:08 i'm using the official alpine docker container, and i'm specifiying a command to run. Is there a way for me to get the output of that command in the docker log? 2022-04-28 18:29:26 alternatively, is there a way to get the container to not exit when its done with whatever its doing? 2022-04-28 18:30:03 entrypoint needs to be non-exiting command like tail -f /dev/null 2022-04-28 18:30:32 alright, i'll put that at the end and see what we come up with from there 2022-04-28 18:30:41 maybe its mad about something else 2022-04-28 18:31:47 and the command needs to output to stdout and/or stderr for the output to be visible in `docker logs` 2022-04-28 18:36:35 Usually you have some application that keeps running 2022-04-28 18:39:12 hmmmm it seems sendmail isn't happy is why its not working, but the same command worked on the iso in a vm 2022-04-28 18:39:44 any error? 2022-04-28 18:39:59 what command you are trying and what are you trying to achieve exactly 2022-04-28 18:40:12 yea, its mad because there's no openssl in the container 2022-04-28 18:40:33 i'm trying to automatically send an email containing an attachment whenever one appears in a specific folder 2022-04-28 18:41:04 `sendmail -f $SENDER_EMAIL -amLOGIN -au$SENDER_EMAIL -ap$SENDER_PASSWORD -H openssl s_client -quiet -tls1 -connect smtp.gmail.com:465 $RECIPIENT_EMAIL` fails 2022-04-28 18:41:45 because the openssl dependency isn't there and sendmail sees the -quiet and other options as an option it should be processing 2022-04-28 18:45:05 did you install it via postfix or exim package? 2022-04-28 18:46:28 or is it using `openssl` cli 2022-04-28 18:46:48 neither. I assumed it was included because it was included on the iso 2022-04-28 18:46:59 the docker image is very minimal 2022-04-28 18:47:04 right 2022-04-28 18:47:17 docker images are bare rootfs 2022-04-28 18:47:26 so the next question is how do i add it without having to compile my own? 2022-04-28 18:47:32 apk add openssl 2022-04-28 18:47:48 is there a more persistent solution? 2022-04-28 18:47:56 build own docker image 2022-04-28 18:47:58 ^ 2022-04-28 18:49:46 something like `printf "FROM alpine:latest\nRUN apk add openssl" | docker build -t my-image -` 2022-04-28 18:50:15 and then run `my-image` instead of `alpine:something` 2022-04-28 18:51:02 but also remember to pull latest `alpine:latest` and rebuild image to have security updates 2022-04-28 18:51:21 s/image/image from time to time 2022-04-28 18:51:21 panekj meant to say: but also remember to pull latest `alpine:latest` and rebuild image from time to time to have security updates 2022-04-28 19:37:36 huzzah! it works! proper emails, with attachments, OCR'd from my scanner with no manual intervention 2022-04-28 19:38:08 \o/ 2022-04-28 20:40:56 wasutton3: on this date skynet became selfware ;) 2022-04-28 20:44:32 Oh no no no no 2022-04-29 02:43:21 Does anyone here know how to install a "diskless" system to an nvme drive? 2022-04-29 02:43:32 I've partition/formatted/setup-bootable on it, but when I boot from it, grub loads, and then I get the following message: 2022-04-29 02:43:42 Mounting boot media failed. 2022-04-29 02:43:43 initramfs emergency recovery shell launched.... 2022-04-29 02:44:58 From what I've read, it seems the nvme module is not available, but I don't understand how to make available. 2022-04-29 02:48:00 edit mkinitfs.conf, add nvme feature 2022-04-29 02:48:11 theoretically it should be automatic but i think it is broken in some cases 2022-04-29 02:51:22 so i tried that...but i'm not clear how to do it for this yet to be configure system...meaning, at this point all i've done is "setup-bootable" and then tried to boot from that 2022-04-29 02:53:01 from a bootable usb, I tried installing mkinitfs, and editing the config file, and then mkinitfs and putting the initfs on the nvme, but either i didn't do something right, or it was the wrong approach 2022-04-29 02:59:16 hm... i think it involves update-kernel somehow 2022-04-29 02:59:31 update-kernel -F nvme 2022-04-29 03:00:38 but i'm not sure how exactly 2022-04-29 03:03:12 ok, i'll look into that....would you run that from a bootable usb you think? 2022-04-29 03:28:48 dunno 2022-04-29 08:11:29 sounds like the problem is that we dont have nvme modules in initramfs by default 2022-04-29 08:13:25 there have been some suggestions to add it by default 2022-04-29 08:14:02 https://gitlab.alpinelinux.org/alpine/alpine-conf/-/issues/10502 2022-04-29 12:02:57 thanks @ikke...do you know how I could add nvme to initramfs manually? This (https://wiki.alpinelinux.org/wiki/NVME) seems to be what I need, but I'm not quite sure how to use it. 2022-04-29 12:03:36 What bootloader is used? 2022-04-29 12:14:27 hello, does this wiki section contain an error? https://wiki.alpinelinux.org/wiki/Create_a_Bootable_Device#Manually_copying_Alpine_files 2022-04-29 12:14:39 I did `sudo mkdosfs -F32 /dev/mmcblk0p1` but it says "mkdosfs: Too few blocks for viable filesystem" 2022-04-29 12:20:06 try mkfs.fat 2022-04-29 12:20:29 but that sounds like.. /dev/mmcblk0p1 being too small? how big is it 2022-04-29 12:20:47 ya, mkdosfs -F32 has always worked for me 2022-04-29 12:21:04 ah, that is symlinked to mkfs.fat anyway 2022-04-29 12:21:10 i would guess the partition is too small 2022-04-29 12:21:12 you can try -F16 2022-04-29 12:21:23 or -F12, not sure how supported they are 2022-04-29 12:22:02 F32 has 32MB as a minimum size i think 2022-04-29 12:22:13 /dev/mmcblk0p1 * 2048 2048 1 512B c W95 FAT32 (LBA) 2022-04-29 12:22:37 your partition is 512bytes in size 2022-04-29 12:22:49 yeah it's way too small, but that's maybe because the sdcard I'm using is a fresh one, though I thought `fdisk -w` wiped everything on there. 2022-04-29 12:23:05 no, it's too small, because it's formatted as 512 bytes 2022-04-29 12:23:11 format it as.. more bytes 2022-04-29 12:23:35 the mkfs tools won't resize it for you, that's all 2022-04-29 12:23:47 yeah so I suppose the instructions on the wiki might be wrong? Because I followed those to a tee. 2022-04-29 12:24:05 I'll try fdisking it manually and see how it goes 2022-04-29 12:24:15 ah, you mean that vertical fdisk block 2022-04-29 12:24:21 ya 2022-04-29 12:26:44 i assume the 2048 part is what breaks it 2022-04-29 12:29:25 ah, if the thing you are formatting has less than 2048 sectors, then putting that in just makes it create a single 512b on the end of the disk.. not sure why that would be the case though 2022-04-29 12:29:43 yeah I think so too: https://paste.debian.net/plain/1239412 2022-04-29 12:30:04 in my case it's a 16G sdcard 2022-04-29 12:32:19 I'm not sure what `-0` means in that context, but I'll just try to make it `+1G` 2022-04-29 12:32:33 ah 2022-04-29 12:32:38 the message doesn't say that negative values are allowed 2022-04-29 12:32:51 at a guess, it only works with busybox fdisk 2022-04-29 12:33:38 util-linux 2.38 for me shows negatives as allowed 2022-04-29 12:33:44 i assume it means to fill the device 2022-04-29 12:33:58 2.31.1 though... idk where that is from 2022-04-29 12:34:19 it's from 2017 2022-04-29 12:34:21 lol 2022-04-29 12:34:30 oh, that's probably it 2022-04-29 12:34:31 yeah I'm on Ubuntu 18.04 2022-04-29 12:34:47 had no idea fdisk was prone to change like that :p 2022-04-29 12:35:19 -0 does not work with busybox fdisk 2022-04-29 12:35:42 or with ancient util-linux apparently 2022-04-29 12:36:06 I wonder in which version this -0 was added, might be a good idea to add a minimum fdisk version to the wiki. 2022-04-29 12:36:27 alternatively it should be replaced with just an enter key press 2022-04-29 12:36:32 because the default is to use the disk 2022-04-29 12:37:59 right, but that would mean to have a single FAT32 (boot) partition on the disk, right? 2022-04-29 12:39:02 that's what it already tries to do 2022-04-29 12:39:17 ah 2022-04-29 12:39:17 and does, with modern util-linux 2022-04-29 12:55:07 Also, `${mydev}1` is not always valid, since in my case it would be `${mydev}p1`, but then again I think the naming scheme is given by systemd and the wiki sections assume the sdcard prep happens on Alpine. 2022-04-29 13:42:35 no, those are kernel names 2022-04-29 13:42:50 they long predate systemd 2022-04-29 13:43:37 udev manages /dev/disk/by-* but alpine has persistent-storage which does roughly the same 2022-04-29 17:51:23 So, how should I install alpine if I want to pass flags to setup-disks? Should I run each setup- script which is normally called by setup-alpine? 2022-04-29 18:33:31 i have upgrade my alpine from 12 to 15, but now my packages aren't loaded from the usb cache any more 2022-04-29 18:33:42 it's booting form usb with lbu for persistency 2022-04-29 18:34:46 i don't get an error, and i tried re-running setup-apkcache multiple times now 2022-04-29 18:34:57 i don't know what triggers whether the cached apks are loaded or not... 2022-04-29 18:42:23 Ermine: https://git.alpinelinux.org/alpine-conf/tree/setup-alpine.in#n226 2022-04-29 18:42:40 horo1: did you upgrade the packages 2022-04-29 18:42:51 yep 2022-04-29 18:43:01 every time anew... 2022-04-29 18:43:33 in any case that's what i meant with upgrading from 12 to 15 2022-04-29 18:44:01 also, it seems that mkinitfs-3.6.0-r0.trigger should depend on cpio? 2022-04-29 18:44:06 i had to install cpio manually 2022-04-29 18:44:26 or in other words, the busybox cpio version doesn't suffice 2022-04-29 18:54:33 i *think* the reason is that the mount changed: 2022-04-29 18:54:53 it's now /dev/sdg1 on /media/sdg1 instead of /media/usb 2022-04-29 18:56:45 the convenient /media/usb is some kind of magic that i never understood, but also never questioned. 2022-04-29 19:00:07 https://gitlab.alpinelinux.org/alpine/aports/-/commit/7141b493a223fdb910ed90439cc044a9d5fe8c46 2022-04-29 19:06:51 so /dev/usbdisk still points to sdg1 2022-04-29 19:07:08 and mount /media/usb works 2022-04-29 19:07:18 (as that is in fstab) 2022-04-29 19:08:20 so i'm not sure if this commit is supposed to have broken anything related to this 2022-04-29 19:09:03 the /dev/disk-by stuff doesn't seem to interfere with the /dev/usbdisk location 2022-04-29 21:14:13 yeah, so i even have modloop and apkvol working, it's just not mounted at the right place, instead at it's drive letter. 2022-04-29 21:14:51 maybe alpine gets confused then and refuses to mount it again at a different place (/media/usb) 2022-04-29 21:15:03 and thus my local apk repo is never found 2022-04-29 21:23:44 thanks minimal 2022-04-29 21:23:52 good lesson to never upgrade to stable lol 2022-04-29 21:25:26 does anyone know why the name alpine was chosen for the project? 2022-04-29 21:27:18 there isn't a good story behind it, unfortunately, though I like to think it's about alpine climbing, where climbers generally take as little as they need 2022-04-29 21:31:04 eddsalkield: ask ncopa? "The codename is Alpine." https://web.archive.org/web/20160514053441/https://osdir.com/ml/linux.leaf.devel/2005-08/msg00039.html 2022-04-29 21:31:51 huh interesting 2022-04-29 21:31:53 c7s: i like this interpretation 2022-04-29 21:34:43 actually I've not seen this mentioned before: ""Alpine" originally stood for A Linux Powered Integrated Network Engine. The idea was that the distribution would be focused on networking, and be a tiny "engine" or framework, which larger systems could be built upon." https://wiki.alpinelinux.org/wiki/Alpine_Linux:Trivia 2022-04-29 21:36:44 Hi everyone, new alpine user here. Coming from debian. 2022-04-29 21:37:57 I was wondering what's up with firefox in alpine stable 3.15. Firefox is still on version 94 in alpine repos, while the latest release is 99. 2022-04-29 21:38:17 Is it recommended to pin firefox to edge? 2022-04-29 21:38:56 Oooh, firefox-esr seems updated to the latest version 91.8. Are we supposed to use esr? 2022-04-29 22:07:42 will installing awall break the rules I've set with ufw? 2022-04-29 22:49:45 FeastOkay: I think it is expected for package versions to remain (mostly) stable in a release branch. I use edge for this reason, but breakage can sometimes occur 2022-04-30 03:02:59 Hello71 and ikke....thanks much for your help with my nvme question, i finally got pieces put together from what you guys said and from these pages https://wiki.alpinelinux.org/wiki/NVME https://www.fedux.org/articles/setup-alpine-linux-diskless.html 2022-04-30 03:05:27 basically....i booted from a usb stick...formatted, setup-bootable on the nvme...installed mkinitfs and squashfs-tools....mounted the nvme to /mnt....ran update-kernel /mnt/boot....rebooted from nvme 2022-04-30 06:41:37 Hi, I'm having a problem with CPU frequency scaling and the Mindustry flatpak. I'm using a Thinkpad X1 Carbon Gen 1 (i5-3317U) with Alpine 3.15, kernel version 5.15.32-0-lts. In order to reduce heat output, I've disabled turbo boost and limited it to 1.6GHz. I did this by setting the intel_pstate driver to 'active' mode, setting '1' to no_turbo, then setting scaling_governor to 2022-04-30 06:41:38 'powersave' and scaling_max_freq to '1600000' in /sys/devices/system/cpu/cpu*/cpufreq/. By checking scaling_cur_freq, I see that the powersave governor is working as expected (scaling between 800MHz which the minimum and 1.6GHz based on load), and the freqeuency doesn't go above 1.6GHz when running multiple instances of 'yes' (to generate load). The problem is that when I run 2022-04-30 06:41:40 the Mindustry flatpak, and click 'Play' then 'Campaign', the frequency goes to around 2.5GHz, although the settings in /sys don't appear to have been changed. Weirdly it isn't increased when in the Mindustry loading screen (before clicking Play), even though during that time the CPU is at around 100% utilization. 2022-04-30 06:43:00 I tried running multiple 'yes' instances in the shell of the Mindustry flatpak (by adding '--command=sh' to the 'flatpak run' command), but even at 100% utilization that didn't cause the CPU frequency to change. 2022-04-30 10:58:46 reducing max freq usually increases power usage 2022-04-30 10:59:57 it is usually more efficient to do the work quickly and then go to sleep 2022-04-30 11:00:52 this might be less true of a game where new frames must be produced every 16ms 2022-04-30 11:01:58 Yes, since it's constant but medium load, reducing the frequency will reduce heat output, right? 2022-04-30 11:02:16 the problem with games is that the work increases to fill the CPU capacity 2022-04-30 11:03:03 What do you mean? 2022-04-30 11:03:15 yes, but i think the kernel should do this automatically if appropriate 2022-04-30 11:04:10 The thing is, it's not going to throttling temperatures, but I'd like it to run slower/cooler. 2022-04-30 11:04:26 I just have no idea why Mindustry somehow is getting the CPU to run faster than scaling_max_freq. 2022-04-30 11:04:37 i mean that if the cpu is slow, the game will do less work ("drop frames"), so the premise of "equal work" is not true 2022-04-30 11:04:55 doesn't mindustry have a frame limiter 2022-04-30 11:07:42 Ah, didn't see that previously. Still, how is it causing the CPU to run faster than set by the governor? 2022-04-30 11:18:41 i don't think this is related to flatpak 2022-04-30 11:18:50 it might be related to gpu 2022-04-30 11:19:20 also, i have heard that undervolting can save power on intel cpus 2022-04-30 11:19:50 i have also heard that undervolting is no longer permitted for security reasons 2022-04-30 11:34:57 Hello, I've recently configured an email server and got a TLS certificate with Let's encrypt via acme-client. The cert seems to be valid. Yet when I try to configure the email client with aerc, it says that the cert is signed by an unknown authority, it happens both on my desktop alpine linux install and my mobile install as well (sxmo/pmOS), yet it doesn't happen with my work 2022-04-30 11:34:59 computer (macOS), so the problem doesn't seem to be with aerc. I've installed ca-certificates with apk and tried to run doas update-ca-certificates, after which I get the following warning: ca-certificates.crt does not contain exactly one certificate or CRT: skipping. Does anyone know where the problem may reside? Most resources point to a mistrusted CA, but I thought Let's 2022-04-30 11:35:01 encrypt is pretty standard 2022-04-30 12:23:15 99% you're missing the intermediate cert 2022-04-30 12:24:04 How would I go to aquire it? 2022-04-30 12:34:31 how did you configure the server 2022-04-30 12:38:19 Sorry, got briefly disconnected 2022-04-30 12:38:32 I use acme-client for ssl cert generation 2022-04-30 12:40:08 can you show `openssl s_client -showcerts -connect yourserver:port` 2022-04-30 12:40:26 that will show what it actually serves 2022-04-30 12:42:05 https://pastebin.com/DXGbvucF 2022-04-30 13:11:53 what does alpine use to manage tmpfile creation? 2022-04-30 13:14:07 mktemp? 2022-04-30 13:15:16 are you referring to tmpfiles.d? 2022-04-30 13:37:19 yes tmpfiles.d 2022-04-30 13:37:37 where is that located please? 2022-04-30 15:05:02 usually stuff under /etc/tmpfiles.d 2022-04-30 15:05:10 idk what triggers it though :) never used it for anything 2022-04-30 15:07:48 mmm I don't have tmpfiles.d which means i don't have anything requiring a tmpfile parser 2022-04-30 15:08:01 I suspect it isn't systemd-tmpfiles for alpine 2022-04-30 15:08:12 maybe opentmpfiles (or a fork) 2022-04-30 15:39:44 isn't tmpfiles.d a systemd thing? 2022-04-30 15:40:06 yes and no 2022-04-30 15:40:34 it does exist outside of systemd world, too 2022-04-30 15:41:58 they created teh problem, made a (flawed) solution. 2022-04-30 15:42:07 opentmpfiles did exist but got abandoned by gentoo 2022-04-30 15:44:07 gentoo just builds systemd tmpfiles now, separately 2022-04-30 15:45:49 yup 2022-04-30 15:46:00 abandoned their project a few months ago 2022-04-30 15:47:13 iirc the abandonment/migration notice and move to systemd-tmpfiles was like 2 years ago 2022-04-30 15:47:59 etmpfiles? 2022-04-30 15:48:05 notice maybe but practically a few months ago 2022-04-30 15:48:38 maybe a year ago (damn how time flies) 2022-04-30 15:48:44 https://github.com/OpenRC/opentmpfiles/issues/19 2022-04-30 19:41:03 Which package contains kernel headers? There's linux-headers package, but this one for 5.16. Also, there is a linux-lts-dev package, but it does not have /usr/include/linux 2022-04-30 19:42:13 https://pkgs.alpinelinux.org/contents?file=*&path=%2Fusr%2Finclude%2Flinux&name=&branch=v3.15&arch=x86_64