#archlinux32 | Logs for 2018-08-15

Back
[00:12:23] -!- bulletmark has joined #archlinux32
[00:18:04] <buildmaster> python-pytest-twisted is broken (says rechenknecht).
[00:56:15] -!- eduardoeae_ has joined #archlinux32
[01:12:34] -!- eduardoeae_ has quit [Ping timeout: 256 seconds]
[01:14:44] -!- eduardoeae_ has joined #archlinux32
[01:25:28] -!- eduardoeae_ has quit [Ping timeout: 260 seconds]
[01:27:05] -!- eduardoeae_ has joined #archlinux32
[01:59:43] -!- Jkx has joined #archlinux32
[01:59:48] <Jkx> Hi guys
[02:00:40] <Jkx> I just updated, but I run into big issues. systemd said it need GLIBC 2.28
[02:01:07] <Jkx> but pacman only find 2.27
[02:02:45] <tyzoid> Jkx: What mirror are you using?
[02:03:10] <Jkx> archlinux32.vollzornbrot.de
[02:03:24] <Jkx> i'm in stable, not testing
[02:03:53] <Jkx> so 2.27 looks fine, I don't understand why systemd need 2.28
[02:05:39] <tyzoid> one sec
[02:07:01] <tyzoid> Looks like it should only need 2.27...
[02:07:39] <tyzoid> damn, need to get deptree running :/
[02:09:51] <tyzoid> one sec, booting my i686 vm
[02:10:03] <tyzoid> Jkx: This'll be about ~10-15 mins to debug
[02:11:15] <Jkx> tyzoid: ok
[02:12:53] -!- Jkx has quit [Quit: WeeChat 2.1]
[02:16:26] -!- Jkx has joined #archlinux32
[02:20:11] <tyzoid> hmm
[02:20:17] <tyzoid> I assume this is what the issue you're hitting is?
[02:20:21] <tyzoid> /usr/lib/initcpio/busybox: /usr/lib/libc.so.6: version `GLIBC_2.28' not found (required by /usr/lib/initcpio/busybox)
[02:20:25] <elibrokeit> tyzoid: is this the first time you all needed to deal with a glibc update?
[02:20:37] <tyzoid> elibrokeit: afaik, I think so
[02:20:44] <elibrokeit> Well, fun
[02:21:18] <elibrokeit> You cannot build for glibc in testing and then move to stable without first moving glibc
[02:21:34] <elibrokeit> Regardless of the soname matching okay
[02:21:59] <tyzoid> elibrokeit: Yeah, though this is something I'll need to bring up with deep42thought
[02:22:07] <tyzoid> since I don't really deal with the build system
[02:22:31] <tyzoid> though it seems that just installing the 2.28 version won't work either
[02:22:55] <elibrokeit> Moving glibc to stable should work fine
[02:23:52] <Jkx> tyzoid: that's it
[02:24:47] <tyzoid> Jkx: I installed glibc from https://archive.archlinux32.org
[02:24:50] <tyzoid> but I'm testing it rn
[02:24:58] <tyzoid> not sure if it'll work with a gui, though
[02:25:03] <tyzoid> I've only got a terminal env atm
[02:25:03] <Jkx> I did the same stuff
[02:25:57] <tyzoid> elibrokeit: Problem is I don't know if some packages still depend on the old version
[02:26:02] <tyzoid> is it mostly backward-compatable?
[02:26:22] <tyzoid> or not at all?
[02:26:28] <Jkx> tyzoid: old software doesn't seem to complaint
[02:35:52] <elibrokeit> tyzoid: glibc has an explicit abi contract
[02:36:05] <elibrokeit> It is intended to *always* be backwards compatible
[02:36:08] <tyzoid> nice
[02:36:38] <elibrokeit> That's the whole reason they use versioned symbols and an unchanging soname
[02:37:15] <elibrokeit> The idea is you can build on older systems and use the results with any more recent version of glibc
[02:37:44] <elibrokeit> This is why prebuilt binaries for foo, are usually built on ancient debian
[02:37:59] <elibrokeit> So they can build with a universal glibc
[02:38:46] <elibrokeit> The issue is exclusively that newer versions of glibc have new symbols , which new programs may use, but older glibc does not provide
[02:39:12] <elibrokeit> Hence almost certain guarantee that its forward compatibility is "hell no"
[02:40:51] <elibrokeit> So essentially there's no reason to hold back glibc in testing except fear that the package is fundamentally borked -- which is the original purpose of testing
[02:41:01] <tyzoid> ok.
[02:41:09] <tyzoid> Just surprised the update logic is broken in that case
[02:41:12] <Jkx> tyzoid: I'm still wondering why this dependency was pushed to stable ?
[02:41:18] <elibrokeit> It's not part of any set except the "I'm first" one
[02:41:33] <tyzoid> Jkx: Our update mechanism is all automatic, instead of manual like Arch has
[02:41:41] <tyzoid> This looks to be a case of buggy logic
[02:41:50] <Jkx> tyzoid: ok.
[02:41:57] <elibrokeit> there are only like 3 or 4 people doing this for i686
[02:42:11] <elibrokeit> and the automation assumes a matching soname is "good enough"
[02:42:16] <Jkx> elibrokeit: yes I know :)
[02:42:22] <elibrokeit> which works basically everywhere except for glibc...
[02:42:45] <tyzoid> 3 or 4 people: Deep42thought (1.5), Abaumann (1.5), Tyzoid (0.4)
[02:42:46] <Jkx> elibrokeit: in fact my archlinux32 was a manjaro-32
[02:42:47] <tyzoid> adds up
[02:43:24] <elibrokeit> the solution would be to special-case glibc, check if it is in testing and if so freeze *everything* except for -any packages
[02:43:53] <elibrokeit> or get fancy and inspect the binaries for GLIBC_2_28 symbols
[02:44:00] <Jkx> tyzoid: take care this bug in a bit severe, cause it affected systemd
[02:44:18] <tyzoid> Jkx: Yup, I'm aware. I have no ability to fix it atm
[02:44:18] <elibrokeit> err, GLIBC_2.28
[02:44:22] <tyzoid> gotta wait for deep42though
[02:44:26] <tyzoid> deep42thought*
[02:44:44] <Jkx> this gonna kill some system, stay tuned ;)
[02:51:55] <Jkx> tyzoid: Thanks for the help / support.
[02:52:10] <Jkx> Bye bye all
[02:52:12] <tyzoid> np, thanks for helping us debug / test / do things
[02:52:15] <tyzoid> Jkx: see ya
[02:52:27] -!- Jkx has quit [Quit: WeeChat 2.1]
[02:54:27] -!- bulletmark has quit [Ping timeout: 240 seconds]
[02:59:21] -!- bulletmark has joined #archlinux32
[03:43:14] -!- chromicant has quit [Quit: ZNC 1.7.1 - https://znc.in]
[05:06:06] -!- bulletmark has quit [Remote host closed the connection]
[06:25:20] -!- deep42thought has joined #archlinux32
[06:25:25] <buildmaster> Hi deep42thought!
[06:25:32] <deep42thought> Hi buildmaster, what's up?
[06:25:35] <buildmaster> up? I'm up for 1 day, 22 hours, 58 minutes, load average: 9.43, 8.69, 7.42
[06:25:56] <deep42thought> elibrokeit: thanks for the info with backwardscompatibility, I'm moving glibc right now
[06:33:01] <deep42thought> I think, the clean solution regarding glibc is to add a dependency on the glibc version which was installed during compilation (this is extractable from the package)
[06:33:39] <deep42thought> e.g. systemd now has a dependency on glibc>=2.28-1.3
[06:50:14] -!- NoobAlice has quit [Quit: Leaving.]
[06:56:18] <elibrokeit> deep42thought: would that be the virtual "link" dependencies listed on your archweb setup?
[07:08:00] <deep42thought> elibrokeit: yes
[07:08:07] <deep42thought> or a totally separate type
[07:08:21] <deep42thought> but I think, it's actually really a "link" dependency
[07:09:04] <deep42thought> dependencies are distinguished by two bits: whether they're relevant for building and whether they're relevant for binary packages
[07:09:28] <deep42thought> which leads to 4 dependency types: make, run, check and link dependencies
[07:09:49] <deep42thought> (where checkdepends is not needed for anything, just tracked for the sake of completeness)
[07:36:25] -!- oaken-source has joined #archlinux32
[08:30:49] -!- deep42thought has quit [Quit: Leaving.]
[08:45:33] -!- abaumann has joined #archlinux32
[08:45:37] <buildmaster> Hi abaumann!
[08:45:59] <abaumann> I recovered two machines (which were still running) by simply upgrading to testing (thus taking in glibc 2.28)
[08:46:16] <abaumann> but moving glibc 2.28 to stable seems to be the best move right now.
[08:46:53] -!- abaumann has quit [Client Quit]
[09:25:05] -!- oaken-source has quit [Ping timeout: 272 seconds]
[10:04:45] -!- oaken-source has joined #archlinux32
[10:11:19] -!- eduardoeae has quit [Ping timeout: 272 seconds]
[10:41:04] -!- oaken-source has quit [Ping timeout: 272 seconds]
[10:55:36] -!- deep42thought has joined #archlinux32
[10:55:37] <buildmaster> Hi deep42thought!
[10:55:52] <deep42thought> abaumann: it should have been moved this morning
[10:57:24] <deep42thought> indeed, it was moved at 4:31 utc: https://packages.archlinux32.org
[10:57:25] <phrik> Title:Arch Linux 32 - glibc 2.28-1.3 (i686) (at packages.archlinux32.org)
[11:48:29] -!- N0rootz has joined #archlinux32
[11:57:35] -!- N0rootz has quit [Ping timeout: 240 seconds]
[12:09:53] -!- sahasatvik__ has joined #archlinux32
[12:16:14] -!- eduardoeae has joined #archlinux32
[13:39:06] -!- N0rootz has joined #archlinux32
[14:00:37] -!- N0rootz has quit [Ping timeout: 245 seconds]
[14:09:06] <buildmaster> python-librabbitmq is broken (says buildknecht2).
[14:17:39] -!- N0rootz has joined #archlinux32
[14:55:08] * buildmaster failed to execute a mysql query - can you have a look at "tmp.mysql-functions.query.stdin.2018-08-15T14:51:08.6qtuA0"?.
[14:56:04] <tyzoid> deep42thought: Server still slow :/
[14:56:12] <deep42thought> tyzoid: which server?
[14:56:14] <deep42thought> yours?
[14:56:16] <tyzoid> yeah
[14:56:20] <tyzoid> srv1
[14:56:22] <deep42thought> buildmaster, too
[14:56:28] <deep42thought> that's, why I was asking ...
[14:56:35] <tyzoid> I'm looking at migrating to nginx as the primary reverse proxy
[14:58:58] <tyzoid> it's likely just that I've got too much stuff on that box, and I'm swapping
[14:59:54] <tyzoid> deep42thought: How do you feel about migrating packages.archlinux32 to srv0 from srv1?
[15:00:03] <tyzoid> would the IP change make a difference?
[15:00:09] <deep42thought> yes, minor
[15:00:16] <deep42thought> just need to change a firewall rule
[15:00:33] <deep42thought> e.g. I'm fine with that
[15:00:47] <tyzoid> sounds good.
[15:00:54] <tyzoid> I'll let you know when I'm ready to switch it over
[15:01:00] <deep42thought> ok
[15:02:33] <deep42thought> firewall rule is added, so you can switch over whenever you like
[15:02:47] <tyzoid> ok, sweet. No config. needed on the client side?
[15:03:03] <deep42thought> we might need to re-bootstrap the database
[15:03:13] <deep42thought> but this depends on how clean mariadb handles the switchover
[15:03:25] <tyzoid> I'm doing a copy of the entire chroot
[15:03:26] <deep42thought> I guess, if we "systemctl stop" it, then it should be fine
[15:03:48] <deep42thought> copy of a live db might break it
[15:03:50] <deep42thought> I think
[15:03:56] <tyzoid> we'll see
[15:04:16] <tyzoid> I don't think it'll have too much of an impact
[15:06:16] <tyzoid> yeah, taking a backup on a slow disk doesn't help the effort xD
[15:40:21] -!- Alina-malina has quit [Ping timeout: 240 seconds]
[15:47:49] * buildmaster resumes sanity.
[15:53:49] -!- rpodgorny has joined #archlinux32
[16:00:27] -!- N0rootz has quit [Ping timeout: 240 seconds]
[16:03:07] -!- Alina-malina has joined #archlinux32
[16:13:32] -!- Cthulu201 has quit [Quit: Nowhere special. I always wanted to go there.]
[16:23:05] -!- Cthulu201 has joined #archlinux32
[16:36:06] <buildmaster> python-celery is broken (says nlopc46).
[16:38:13] <tyzoid> wb Cthulu201
[16:41:39] <buildmaster> python-cryptography is broken (says buildknecht2).
[16:44:18] <buildmaster> mate-panel is broken (says nlopc46).
[16:45:28] <buildmaster> python-aiohttp is broken (says rechenknecht).
[16:48:29] <buildmaster> mate-terminal is broken (says buildknecht2).
[16:48:40] <buildmaster> mate-session-manager is broken (says nlopc46).
[16:50:59] <buildmaster> engrampa is broken (says rechenknecht).
[16:51:21] <buildmaster> python-netcdf4-openmpi is broken (says nlopc46).
[16:56:09] <buildmaster> virtualbox-modules-arch is broken (says buildknecht2).
[17:00:56] -!- eduardoeae__ has joined #archlinux32
[17:02:53] -!- eduardoeae_ has quit [Ping timeout: 244 seconds]
[17:03:52] <buildmaster> virtualbox is broken (says nlopc46).
[17:08:25] <buildmaster> python-hypothesis is broken (says buildknecht2).
[17:21:34] <buildmaster> python-zeep is broken (says nlopc46).
[17:23:38] -!- isacdaavid has joined #archlinux32
[17:44:02] <buildmaster> python-wsproto is broken (says rechenknecht).
[17:44:59] <buildmaster> streamlink is broken (says buildknecht2).
[17:46:38] -!- isacdaavid has quit [Quit: Leaving.]
[18:04:23] <buildmaster> python-testresources is broken (says buildknecht2).
[18:17:58] -!- NoobAlice has joined #archlinux32
[18:25:22] <buildmaster> python-mox3 is broken (says eurobuild3).
[18:25:32] <buildmaster> python-oslo-context is broken (says nlopc46).
[18:25:47] <buildmaster> vulkan-docs is broken (says rechenknecht).
[18:45:33] -!- abaumann has joined #archlinux32
[18:45:34] <buildmaster> Hi abaumann!
[18:45:37] <tyzoid> wb
[18:45:50] <abaumann> hi :-)
[18:46:11] <abaumann> deep42thought: all machines up again with glibc 2.28. the euroalix is running now on the i486 branch due to illegal instructions.
[18:47:05] <abaumann> depencies around icu also break currently in Archlinux ARM (for instance with php7)
[18:47:30] -!- oaken-source has joined #archlinux32
[18:50:58] <deep42thought> is our icu built in the meantime?
[18:51:12] <buildmaster> sonic-visualiser is broken (says buildknecht2).
[18:51:14] <abaumann> no.
[18:51:18] <abaumann> I didn't patch yet.
[18:51:29] <abaumann> will do now.
[18:51:36] <abaumann> sorry. was busy recovering my alix. :-)
[18:51:38] <deep42thought> thanks
[18:51:49] <deep42thought> I'm currently a little fed up with python-*
[18:51:50] <deep42thought> :-D
[18:52:17] <abaumann> I can totally understand..
[18:53:55] <buildmaster> vulkan-tools is broken (says rechenknecht).
[18:56:59] <buildmaster> python-hypothesis is broken (says eurobuild3).
[18:58:13] <abaumann> ok. icu fails because the integer parser has been rewritten and apparently not been tested on 32-bit Intel. :-)
[18:58:26] <deep42thought> nice job
[18:59:43] <abaumann> we will have a lot of fun with glibc 2.28 too. so far freadahead and readdir_r are no longer in the api, of course people have been warned 2 years that those things will change, but nobody adapts the code. So gnulib, openjdk fail (so far), more to come..
[19:00:08] <abaumann> seen, upstream also has fun with m4/gnulib. :->
[19:11:16] -!- abaumann has quit [Quit: leaving]
[19:39:43] -!- N0rootz has joined #archlinux32
[19:51:51] -!- N0rootz has quit [Ping timeout: 240 seconds]
[19:53:15] <buildmaster> open-vm-tools is broken (says rechenknecht).
[20:05:04] <buildmaster> yaz is broken (says rechenknecht).
[20:12:45] -!- N0rootz has joined #archlinux32
[20:18:26] -!- rpodgorny has parted #archlinux32
[20:24:08] -!- NoobAlice has quit [Ping timeout: 260 seconds]
[20:26:04] -!- NoobAlice has joined #archlinux32
[20:27:11] <buildmaster> libical is broken (says nlopc46).
[20:28:20] <buildmaster> raptor is broken (says nlopc46).
[20:30:59] <buildmaster> python-pycuda is broken (says nlopc43).
[20:34:56] -!- N0rootz has quit []
[20:41:58] <buildmaster> libfbclient is broken (says rechenknecht).
[20:44:28] <buildmaster> imagemagick is broken (says rechenknecht).
[20:48:43] <buildmaster> io is broken (says buildknecht2).
[21:03:14] <buildmaster> openvdb is broken (says nlopc43).
[21:11:46] <buildmaster> python-mamba is broken (says nlopc46).
[21:19:03] -!- deep42thought has quit [Quit: Leaving.]
[21:26:37] <buildmaster> python-pipenv is broken (says nlopc43).
[21:34:35] <elibrokeit> abaumann, deep42thought: we've seen a couple issues from gnulib, but those are per-package issues... because gnulib is an insane thing that vendors source code into other projects like some sort of horrid Windows software
[21:53:04] <buildmaster> 0ad is broken (says nlopc46).
[22:02:43] -!- oaken-source has quit [Ping timeout: 260 seconds]
[22:11:57] <buildmaster> aegisub is broken (says eurobuild3).
[22:12:30] <buildmaster> widelands is broken (says nlopc46).
[22:20:30] <buildmaster> cantor is broken (says nlopc46).
[22:24:32] <buildmaster> qt5-location is broken (says nlopc43).
[22:27:22] <buildmaster> dnssec-tools is broken (says nlopc46).
[22:28:37] <buildmaster> qemu is broken (says rechenknecht).
[22:28:46] <buildmaster> libsoup is broken (says nlopc46).
[22:28:57] <buildmaster> imagemagick6 is broken (says nlopc46).
[22:39:03] <buildmaster> wine-staging-nine is broken (says nlopc43).
[23:02:45] <buildmaster> maxima is broken (says nlopc43).
[23:05:30] <buildmaster> glom is broken (says eurobuild3).
[23:10:26] <buildmaster> webkit2gtk is broken (says nlopc46).
[23:24:18] <buildmaster> chromium is broken (says rechenknecht).
[23:30:30] <elibrokeit> tyzoid: https://pool.mirror.archlinux32.org is full of errors
[23:30:44] <elibrokeit> > pool.mirror.archlinux32.org uses an invalid security certificate. The certificate is only valid for srv1.tyzoid.com.
[23:31:14] <elibrokeit> <tyzoid> deep42thought: How do you feel about migrating packages.archlinux32 to srv0 from srv1?
[23:31:14] <elibrokeit> <tyzoid> would the IP change make a difference?
[23:31:14] <elibrokeit> <deep42thought> yes, minor
[23:31:14] <elibrokeit> <deep42thought> just need to change a firewall rule
[23:31:18] * elibrokeit lols
[23:43:22] <buildmaster> gitlab is broken (says nlopc46).
[23:57:53] <buildmaster> vulkan-lunarg-tools is broken (says eurobuild3).