#archlinux32 | Logs for 2019-02-15

[00:00:46] -!- thePiGrepper has joined #archlinux32
[00:11:46] -!- guys has quit [Ping timeout: 268 seconds]
[00:38:47] -!- thePiGrepper has quit [Ping timeout: 240 seconds]
[01:19:00] -!- woshty has quit [Ping timeout: 272 seconds]
[01:24:12] -!- samantaz has quit [Read error: Connection reset by peer]
[02:35:18] -!- bill-auger has quit [Ping timeout: 246 seconds]
[02:36:39] -!- bill-auger has joined #archlinux32
[03:16:35] <buildmaster> i686/kde-cli-tools are broken (says eurobuild3).
[03:35:53] <buildmaster> girls, please have a look at my dirty database
[03:35:53] * buildmaster goes insane.
[03:41:44] -!- oaken-source has quit [Ping timeout: 250 seconds]
[04:17:59] * buildmaster resumes sanity.
[05:08:33] -!- thePiGrepper has joined #archlinux32
[06:24:49] -!- deep42thought has joined #archlinux32
[06:24:49] <buildmaster> Hi deep42thought!
[06:24:49] <buildmaster> !rq deep42thought
[06:24:50] <phrik> buildmaster: <deep42thought> I forgot to copy over the login credentials
[06:25:01] <deep42thought> abaumann: /usr/bin/systemd-tmpfiles: error while loading shared libraries: libidn2.so.4: cannot open shared object file: No such file or directory
[06:25:05] <deep42thought> ^on i486
[06:25:26] <deep42thought> looks, like something is not (yet) right :-)
[06:46:05] <elibrokeit> !moar rebuilds
[06:46:05] <phrik> moaarrr rebuilds!!! MMMOOOOOAAAAAARRRRRRR!!!
[06:55:58] <deep42thought> lol
[06:56:32] <deep42thought> !moar quotes
[06:56:32] <phrik> moaarrr quotes!!! MMMOOOOOAAAAAARRRRRRR!!!
[07:00:35] <deep42thought> buildmaster: what's up?
[07:01:06] <buildmaster> up? I'm up for 1 day, 49 minutes, load average: 1.24, 1.75, 1.64
[07:01:54] -!- deep42thought has quit [Quit: Leaving.]
[07:34:10] <bill-auger> !moar cowbell
[07:34:11] <phrik> moaarrr cowbell!!! MMMOOOOOAAAAAARRRRRRR!!!
[08:05:26] -!- abaumann has joined #archlinux32
[08:05:26] <buildmaster> Hi abaumann!
[08:05:27] <buildmaster> !rq abaumann
[08:05:27] <phrik> buildmaster: <abaumann> it usually automatically fails. :-)
[08:06:51] <abaumann> deep42thought: yeah. no matter how I build, I get a dependency on libidn2.so.4. The problem is: systemd always depend on libidn2.so.4, becuase I have to keep on the virtual machine for the old systemd not to have link errors, installing libidn2.so.0 then gives a libidn2.so pointer to libidn2.so.0, but systemd links against libidn2.so.4..
[08:29:55] -!- deep42thought has joined #archlinux32
[08:29:55] <buildmaster> Hi deep42thought!
[08:29:55] <buildmaster> !rq deep42thought
[08:29:56] <phrik> buildmaster: <deep42thought> one day, some genius will prove, that all programming languages are isomorphic and there is actually only _one_ compiler
[08:30:37] <deep42thought> abaumann: hmm, maybe it links randomly?
[08:30:45] <deep42thought> because I think, we did exactly this way on i686
[08:31:05] <deep42thought> we had libidn2.so.4 installed (as shim, I guess) and it magically linked against libidn2.so.0
[08:40:00] <abaumann> hi deep42thought
[08:40:13] <deep42thought> Hi abaumann
[08:40:28] <abaumann> yeah, that's how I understand a linker works. You link against the thing linked to by the libidn2.so symlink
[08:40:57] <deep42thought> and this points to the .0?
[08:41:33] <abaumann> yeah. Thought so. Anyway. I think my build slave was severly borked. I'll start from scratch..
[08:42:22] <abaumann> the shim contained a copy of the original library, I suppose..
[08:42:29] <deep42thought> yes, it does
[08:42:32] <abaumann> ok.
[08:42:47] <deep42thought> I should really create a shortcut bash script for that
[08:43:14] <deep42thought> a.k.a. 'create-build-support-package-from i686/libidn2-$pkgver-$pkgrel $file1 $file2 $file3 ...'
[08:43:49] <abaumann> TODO list :-)
[08:45:57] <abaumann> you didn't have problems with pacman and gnutls also using libidn2?
[08:46:14] <abaumann> and gnupg?
[08:46:20] <deep42thought> not that I noticed
[08:46:25] <abaumann> mmh..
[08:46:43] <deep42thought> but tbh, bs1 is dead, currently and only bs0 works ...
[08:55:41] <deep42thought> pacman: error while loading shared libraries: libidn2.so.4: cannot open shared object file: No such file or directory
[08:55:45] <deep42thought> ah, you're right :-)
[08:59:18] <abaumann> I think, it's transitive via libpsl and gnupg
[08:59:47] <abaumann> so we should rebuild the reverse tree of patree -r libidn2
[08:59:50] <abaumann> *pactree
[09:00:45] <deep42thought> or we identify the one or two bad packages ;-)
[09:01:06] <abaumann> iputils
[09:01:10] <abaumann> libpsl
[09:01:16] <abaumann> gnutls
[09:01:18] <abaumann> bind-tools
[09:01:24] <abaumann> mutt
[09:01:26] <abaumann> s-nail
[09:01:35] <abaumann> wget
[09:01:36] <abaumann> whois
[09:01:54] <abaumann> only pacman and systemd is really critical
[09:04:09] <abaumann> they seem all fine on i686
[09:04:28] <abaumann> so, they will build on i486 when I fixed the systemd issue and continue normal building..
[09:05:00] <deep42thought> yeah
[09:05:06] <deep42thought> at least, that worked on i686 ;-)
[09:32:00] -!- woshty has joined #archlinux32
[09:37:52] <abaumann> yep. It works. systemd is ok for i486, now to pacman. Then build slaves should recover..
[09:38:05] <deep42thought> "recover"? how so?
[09:38:26] <abaumann> well.. deleting /var/lib/archbuild and installing from staging currently fails. :-)
[09:39:05] <deep42thought> ah, ok, yes - if you didn't update to the faulty systemd/pacman/libidn2 version
[09:39:08] <deep42thought> on the host^
[09:39:13] <deep42thought> ... which I did ...
[09:39:20] <abaumann> ah.
[09:39:24] <abaumann> true. forgot about the host.
[09:41:34] <deep42thought> I should really check if the build environments have severe issues befor updating the host
[09:47:04] -!- oaken-source has joined #archlinux32
[09:58:13] <abaumann> ldd /usr/bin/pacman | grep libidn2 libidn2.so.0 => /usr/lib/libidn2.so.0 (0xb7741000) libidn2.so.4 => /usr/lib/libidn2.so.4 (0xb7312000)
[09:58:16] <abaumann> nice :-)
[09:58:21] <deep42thought> \o/
[09:58:28] <abaumann> ahem.
[09:58:41] <deep42thought> err, wait
[09:58:46] <abaumann> :-)
[09:58:47] <deep42thought> so it links agains _both_ ???
[09:58:49] <abaumann> yes.
[09:58:52] <deep42thought> loool
[09:58:58] <abaumann> one is: libnghttp2.so
[09:59:17] <deep42thought> use lddtree and find out the real lib which links against the wrong one, then
[09:59:24] <abaumann> aeh. sorry: libcurl.so.4 -> libidn2.so.0
[09:59:26] <deep42thought> unfortunately, I cannot help, as my pacman is broken
[09:59:33] <deep42thought> and lddtree is not installed :-(
[09:59:33] <abaumann> libpsl.so.5 -> libidn2.so.4
[09:59:38] <abaumann> no problem.
[09:59:57] <deep42thought> ... and eli does not provide pacman-static binaries for i486 ;-P
[09:59:57] <abaumann> I'm checking in parallel on my i686 staging environment, if everything is fine too.
[10:00:26] <abaumann> pax-utils is available for i486 :-)
[10:00:54] <deep42thought> yes, but I need a working pacman to install it
[10:01:03] <deep42thought> .. or pacman-static ;-)
[10:01:04] <abaumann> oh. :-)
[10:01:37] <abaumann> I don't remember where I got stuck completly with pacman-static.. I think it was in musl somewhere..
[10:01:45] <abaumann> didn't eli make it work?
[10:02:01] <deep42thought> on i686, yes
[10:02:05] <abaumann> ah. ok.
[10:02:21] <deep42thought> but maybe this means, it works on i486, too?
[10:03:27] <abaumann> I'll try again..
[10:13:57] <abaumann> mmh. my build host for i486 is on staging. that's also somewhat risky..
[10:14:08] <deep42thought> :-D mine are too, I think
[10:14:21] <deep42thought> yup
[10:14:40] <abaumann> maybe this is because there was no stable when we started?
[10:14:54] <deep42thought> well
[10:15:01] <deep42thought> it doesn't make sense for the _host_
[10:15:07] <deep42thought> the chroots should be on staging
[10:15:10] <deep42thought> but not the host
[10:15:14] <abaumann> exactly
[10:15:25] <abaumann> I'll change that on my build slave and do a downgrade.
[10:15:42] <deep42thought> ah, I know
[10:15:57] <deep42thought> I was building directly on the build slave (w/o systemd-nspawn)
[10:16:06] <deep42thought> to test some stuff and bootstrap
[10:16:07] <abaumann> ah.
[10:19:50] <deep42thought> good news: i486 on testing is not (yet?) broken :-D
[10:21:15] <abaumann> it will be (soon) :-)
[10:21:22] <deep42thought> :-D
[10:21:50] <abaumann> oh. 64MB is not enough for the shutdown ramfs, I get an OOM while shutting down my i486 test system.
[10:22:11] <abaumann> System is deadlocken on memory.. cool.
[10:22:22] <abaumann> never seen that message in the Linux kernel..
[10:22:49] <deep42thought> what's the bailout for OOM - shutting down?
[10:23:02] <abaumann> yeah. VM forced power off.
[10:23:18] <abaumann> OTOH: I'm pretty impressed that systemd runs with 64MB RAM on a i486..
[10:23:47] <deep42thought> hmm, bs1 has a kernel panic
[10:24:18] <deep42thought> ah, the initramfs is damaged
[10:24:20] <deep42thought> good
[10:25:26] <deep42thought> err, how do I repair that? can I simply copy over the initramfs from the other build slave?
[10:25:47] <abaumann> Yeah, I think so.
[10:26:04] <abaumann> They don't differ in package versions and hardware hooks.
[10:26:31] <abaumann> the OOM was when I replaced systemd.. now it behaves ok.
[10:28:17] <abaumann> ping
[10:28:18] <abaumann> ping: error while loading shared libraries: libidn2.so.4:
[10:28:32] <abaumann> yeah iputils..
[10:40:35] <abaumann> pacman -Qo /usr/lib/libidn2.so.0 and lrwxrwxrwx 1 root root 12 Feb 15 10:40 libidn2.so.0 -> libidn2.so.4
[10:40:39] <abaumann> what the d***
[10:40:48] <deep42thought> :-D
[10:40:54] <deep42thought> is this some kind of workaround?
[10:42:24] <abaumann> mmh.
[10:44:25] <abaumann> mmh, pacman does weird stuff when installing symlinks.
[10:44:36] <abaumann> shouldn't it delete the symlink before installing the new one?
[10:44:56] <abaumann> I remember reading something on the pacman-dev list
[10:48:55] <abaumann> /usr/lib/systemd/systemd-resolved: error while loading shared libraries: libidn2.so.4: cannot open shared object file: No such file or directory
[10:49:04] <abaumann> mmh. something is still linking to the wrong library..
[10:50:43] <deep42thought> systemd: the fairy tale of the program which tried to do everything
[10:52:48] <abaumann> technically, it's a set of programs..
[10:53:00] <deep42thought> until now, it is
[10:53:11] <deep42thought> still: if you break systemd _nothing_ will work
[10:53:17] <deep42thought> not even resolving names ...
[10:53:30] <abaumann> if you use systemd-resolved, yes.
[10:53:34] <abaumann> ldd /usr/lib/systemd/systemd-resolved | grep idn2 libidn2.so.0 => /usr/lib/libidn2.so.0 (0xb7a59000) libidn2.so.4 => not found
[10:53:36] <deep42thought> which is the default
[10:53:41] <abaumann> that's a weird one.
[10:53:49] <abaumann> another library creeping in, indirectly.
[10:54:09] <abaumann> the problem with ldd/lddtree is: if the program uses dlopen, then you don't catch it this way.
[10:54:19] <abaumann> gogo-gadgeto-strace
[10:54:27] <deep42thought> !grab abaumann
[10:54:28] <phrik> deep42thought: Tada!
[10:59:48] <abaumann> mmh. only systemd-resolved seems to be linked against both versions of libidn2. So either it got two dependencies while building in the shimmed chroot (so I can just rebuild systemd once more without the shim) or another library uses libidn.so.4..
[11:00:42] <deep42thought> check lddtree
[11:01:05] <deep42thought> my guess is, that it's a different library which links against .4
[11:01:54] <abaumann> no
[11:01:57] <abaumann> wait. I paste it..
[11:03:10] <abaumann> oh.
[11:03:12] <abaumann> libgnutls.so.30
[11:03:27] <abaumann> why a resolved needs an SSL library, well..
[11:03:34] <deep42thought> doh?
[11:04:15] <abaumann> http://archlinux32.andreasbaumann.cc
[11:04:40] <deep42thought> yup
[11:04:59] <abaumann> so. I'll reschedule gnutls and systemd :-)
[11:05:08] <deep42thought> no
[11:05:10] <deep42thought> only gnutls
[11:05:13] <abaumann> ah. yes.
[11:05:18] <abaumann> true
[11:07:02] <abaumann> libgnutls.so.30 has no dependency on libidn on i686. strange
[11:07:16] <abaumann> *libidn2
[11:07:17] <deep42thought> but it does link against it?
[11:07:25] <abaumann> no, it doesn't seem like.
[11:07:44] <deep42thought> libidn2.so.0 => /usr/lib/libidn2.so.0
[11:07:48] <deep42thought> it does on x86_64
[11:08:04] <deep42thought> and it does on i686
[11:08:07] <deep42thought> (staging)
[11:08:56] <abaumann> lddtree /usr/lib/systemd/systemd-resolved on staging doesn't show it (on my staging slave)
[11:09:12] <abaumann> ah. well.
[11:09:30] <deep42thought> it does on mine
[11:09:45] <deep42thought> ah, wait, it was not up-to-date
[11:10:13] <abaumann> so, maybe they dropped the dependency on libidn2 in gnutls.. checking
[11:10:51] <deep42thought> nope, the updated one still links against libidn2.so.0
[11:10:58] <abaumann> asp32 difflog gnutls
[11:10:58] <abaumann> ==> There is no Arch32 patch for gnutls
[11:11:07] <abaumann> mmh. something is also borked there..
[11:11:18] <deep42thought> -u
[11:11:33] <deep42thought> it's a feature, not a bug
[11:11:57] <abaumann> ah.
[11:12:18] <abaumann> deeviates from the syntax from upstream asp
[11:12:28] <deep42thought> well
[11:12:30] <deep42thought> not really
[11:12:35] <deep42thought> it shows you the log
[11:12:38] <deep42thought> but _our_ log
[11:13:04] <deep42thought> and if you want to see the (less important *cough*) upstream log, you must tell it to "ignore2 archlinux32
[11:13:28] <deep42thought> but I'm open to suggestions about changing the behaviour of -u
[11:13:48] <abaumann> upstream gnutls added an explicit libidn2.so dependency
[11:14:19] <deep42thought> elibrokeit said, they added those dependencies to quite everything to avoid breakage
[11:14:35] <deep42thought> (I mean "everything which links against libidn2")
[11:14:35] <abaumann> yep. sounds reasonable.
[11:24:21] <abaumann> https://archlinux32.org
[11:24:22] <phrik> Title: Arch Linux 32 - List of Build Slaves (at archlinux32.org)
[11:24:48] <abaumann> something doens't work with those forced-package-build scripts. Now nlopc46-i486bs0 and tyzoid-srv0-vm486 are both building i486/gnutls
[11:26:07] <abaumann> yeah. asp32 -u is ok. I don't see another option. It's logical for asp32 do show the diffs of our log.
[11:26:14] <abaumann> asp32 -u shows the ones from upstream
[11:26:22] <abaumann> for difflog, I mean.
[12:00:09] <abaumann> seems everything is ok in the i486 world now, at least to build packages again..
[12:03:17] <bill-auger> the ones that got me were gnutls libpsl and curl
[12:06:48] <abaumann> I add wget to the list :-)
[12:08:03] <bill-auger> oh yea there are more - those were in a minimal chroot
[12:41:13] <deep42thought> abaumann: forced-package-builds-* overwrites almost every other logic: the only check that is left is, that the respective package is actually on the build list
[12:41:48] <deep42thought> if you just want to prioritize a build order (which could be built anyway, but might be scheduled later), use prioritize-build-list
[13:09:20] <abaumann> ah. ok.
[13:34:27] -!- Piraty has quit [Remote host closed the connection]
[13:34:56] -!- Piraty has joined #archlinux32
[13:38:14] <buildmaster> i686/python-hypothesis are broken (says nlopc46).
[13:40:46] <buildmaster> i686/alot is broken (says rechenknecht).
[13:50:37] -!- lodger has joined #archlinux32
[14:12:09] -!- thePiGrepper has quit [Read error: Connection reset by peer]
[14:14:44] -!- thePiGrepper has joined #archlinux32
[14:35:43] -!- lodger has quit [Quit: Leaving]
[15:01:47] -!- oaken-source has quit [Ping timeout: 240 seconds]
[15:40:46] -!- deep42thought has quit [Quit: Leaving.]
[15:44:40] -!- abaumann has quit [Quit: leaving]
[16:13:20] -!- Dimtree has quit [Quit: Peace]
[16:38:12] -!- abaumann has joined #archlinux32
[16:38:12] <buildmaster> Hi abaumann!
[16:38:12] <buildmaster> !rq abaumann
[16:38:13] <phrik> buildmaster: <abaumann> * abaumann looks at his hands.. counts them.. shakes his head and sighs about the fact he cannot grow some more..
[16:39:04] -!- Dimtree has joined #archlinux32
[16:39:36] <abaumann> deep42thought: this will be a dangerous transition of libidn2 stuff, I just almost killed all mu i486 machines. The problem is, you need both libidn2 at a certain point. I think, I force everything for i486 now to stable..
[16:41:52] -!- abaumann has quit [Client Quit]
[17:02:24] <elibrokeit> abaumann: cp some files into the build chroot?
[17:06:00] <buildmaster> i686/gitea is broken (says nlopc43).
[17:06:15] <buildmaster> i686/texmacs are broken (says nlopc46).
[17:29:34] -!- abaumann has joined #archlinux32
[17:29:35] <buildmaster> Hi abaumann!
[17:29:35] <buildmaster> !rq abaumann
[17:29:35] <phrik> buildmaster: <abaumann> I remember the old saying in archlinux: "core" is for the things we need to boot the system and build the system. So meson is now part of core?
[17:30:12] <abaumann> elibrokeit: the really dangerous thing to do is to make a symlink like libidn.so.0 to libidn.so.4 or vise versa. This means, when installing libidn2 with pacman, all kind of fun happens.
[17:30:24] <abaumann> So the proper way is _COPY_ or a shim package.
[17:31:58] <abaumann> I was just testing to force all libidn2.0 related packages to stable for i486 (as not many people use it), because at the moment i686 libidn2.s0.0 refuses to go into stable because some packages are still not rebuilt presumably..
[17:33:51] <elibrokeit> Classic bootstrap solution, cp needed files into the build chroot
[17:41:55] <abaumann> mmh. after upgrading some i486 machines I get a symlink from libidn2.so.0 -> libidn2.so.4
[17:42:08] <abaumann> I hope this is just because I did something stupid to them..
[17:42:17] <abaumann> but 5 times the same stupid thing? :-)
[17:45:19] <abaumann> BTW. pam is a real mess ATM
[17:45:24] -!- guys has joined #archlinux32
[17:46:48] <abaumann> oh cool. setting the root partition to read only leads to 'Authentication failure'
[17:46:58] <abaumann> So, something _IS_ generating pam rules on the fly.
[17:47:16] <abaumann> This is really not good for applicances with a root filesystem on a compact flash :->
[17:48:21] <elibrokeit> what is getting auth failures from pam?
[17:48:30] <abaumann> logging in on the text console.
[17:48:42] <elibrokeit> you're aware that the "other" rule was recently changed?
[17:49:01] <elibrokeit> https://bugs.archlinux.org
[17:49:02] <phrik> Title: FS#48650 : [pambase] "other" template is non-upstream (at bugs.archlinux.org)
[17:49:16] <abaumann> yes. it broke tons of things.
[17:49:26] <abaumann> pam_tally2(login:auth): Error opening /var/log/tallylog for update: Read-only file system
[17:49:29] <abaumann> aha.
[17:49:54] <abaumann> My /var/log is on the root device. funny.
[17:50:06] <elibrokeit> indeed, we've got lots of new bugs for pam issues :/
[17:50:28] <elibrokeit> ha, /var on readonly rootfs is not right
[17:50:31] <abaumann> don't get me wrong. the xlock fall-thru automatic login was a bad thing.
[17:50:55] <abaumann> so having a default deny pam rule and adding an xlock pam file is the proper way to go - from the poing of view of proper security
[17:51:13] <abaumann> yeah. read-only root is a corner case, and might need customized pam rules..
[17:51:24] <elibrokeit> especially since upstream xlockmore has pam rules :/
[17:53:32] <abaumann> yeah. removing pam_tally2.so helped. :-)
[18:00:12] <buildmaster> i686/tpm2-abrmd is broken (says nlopc46).
[18:01:48] <abaumann> yep, /var on read-only is very wrong.. I thought I changed that to a tmpfs..
[18:07:22] -!- girls has quit [Ping timeout: 244 seconds]
[18:07:53] -!- tyzoid has quit [Ping timeout: 244 seconds]
[18:14:36] -!- girls has joined #archlinux32
[18:14:36] <buildmaster> Hi girls!
[18:14:36] <buildmaster> !rq girls
[18:14:36] <phrik> buildmaster: <girls> I'm married, I don't cook
[18:15:20] -!- tyzoid has joined #archlinux32
[18:15:20] <buildmaster> Hi tyzoid!
[18:15:20] <buildmaster> !rq tyzoid
[18:15:21] <phrik> buildmaster: <tyzoid> i ned halp geting antergose instaled. any1 halp plox?
[18:23:54] -!- samantaz has joined #archlinux32
[18:34:33] -!- abaumann has quit [Quit: leaving]
[18:42:56] -!- yans has joined #archlinux32
[19:23:10] -!- deep42thought has joined #archlinux32
[19:23:10] <buildmaster> Hi deep42thought!
[19:23:10] <buildmaster> !rq deep42thought
[19:23:12] <phrik> buildmaster: <deep42thought> computers ranked by dirt: smokers, car mechanics, bedroom, ...
[19:35:57] <buildmaster> i686/freealut is broken (says nlopc46).
[19:36:33] <buildmaster> i686/libdmapsharing is broken (says nlopc46).
[19:37:07] <buildmaster> i686/ossp is broken (says nlopc46).
[19:37:45] <buildmaster> i686/harfbuzz is broken (says nlopc46).
[19:38:16] <buildmaster> i686/perl-alien-sdl is broken (says nlopc46).
[19:38:52] <buildmaster> i686/speexdsp is broken (says nlopc46).
[19:40:26] <buildmaster> i686/gpaste is broken (says nlopc46).
[19:41:46] <buildmaster> i686/xdg-user-dirs are broken (says nlopc46).
[19:42:20] <buildmaster> i686/python-mysql-connector is broken (says nlopc46).
[19:42:50] <buildmaster> i686/perl-file-which is broken (says nlopc46).
[19:43:28] <buildmaster> i686/meld is broken (says nlopc46).
[19:45:28] -!- buildmaster has quit [Remote host closed the connection]
[19:45:40] -!- buildmaster has joined #archlinux32
[20:32:31] -!- ofara_ has joined #archlinux32
[21:17:04] -!- samantaz has quit [Ping timeout: 258 seconds]
[21:52:08] -!- samantaz has joined #archlinux32
[23:02:14] <deep42thought> abaumann: If you need a shim-package next time, have a look at create-build-support-package
[23:07:34] -!- deep42thought has quit [Quit: Leaving.]
[23:27:22] -!- autofsckk has quit [Remote host closed the connection]
[23:32:09] * buildmaster goes insane.
[23:55:39] -!- samantaz has quit [Ping timeout: 268 seconds]