#archlinux32 | Logs for 2020-02-13

Back
[00:04:09] <buildmaster> i486/linux is broken (says nlopc46-i486bs1): https://archlinux32.org
[00:19:35] <buildmaster> i486/vis are broken (says nlopc46-i486bs1): https://archlinux32.org
[04:05:52] <buildmaster> pentium4/mame is broken (says nlopc46): https://archlinux32.org
[04:57:20] <buildmaster> pentium4/octave is broken (says rechenknecht): https://archlinux32.org
[05:14:31] -!- skydrome has quit [Ping timeout: 268 seconds]
[05:54:49] <buildmaster> pentium4/gmic is broken (says rechenknecht): https://archlinux32.org
[06:01:00] <buildmaster> pentium4/purpose is broken (says nlopc46): https://archlinux32.org
[06:03:42] <buildmaster> pentium4/flashplugin is broken (says nlopc46): https://archlinux32.org
[06:17:20] -!- bill-auger_ has joined #archlinux32
[06:20:03] -!- bill-auger has quit [Ping timeout: 260 seconds]
[06:24:10] -!- girls has quit [Quit: ZNC 1.7.5 - https://znc.in]
[06:25:46] -!- girls has joined #archlinux32
[06:25:46] <buildmaster> Hi girls!
[06:25:46] <buildmaster> !rq girls
[06:25:47] <phrik> buildmaster: <girls> I'm married, I don't cook
[06:31:53] -!- titus_livius has joined #archlinux32
[06:36:17] <trotz> 2020/02/13 06:35 OK jeti100 PING PING OK - Packet loss = 0%, RTA = 13.51 ms
[06:36:17] <trotz> 2020/02/13 06:35 OK jeti100 SSH SSH OK - OpenSSH_8.1 (protocol 2.0)
[07:18:14] -!- abaumann has joined #archlinux32
[07:18:15] <buildmaster> Hi abaumann!
[07:18:15] <buildmaster> !rq abaumann
[07:18:16] <phrik> buildmaster: <abaumann> so, the next compression algorithm in pacman will require 1 TB of RAM? ;-)
[07:23:50] <buildmaster> pentium4/acpi_call is broken (says nlopc46): https://archlinux32.org
[07:39:20] <trotz> 2020/02/13 07:38 CRIT buildmaster Forum HTTP CRITICAL: HTTP/1.1 503 Service Unavailable - string 'Pacman' not found on 'http://bbs.archlinux32.org:80/' - 2875 bytes in 0.037 second response time
[07:39:21] <trotz> 2020/02/13 07:38 CRIT buildmaster Forum Secure HTTP CRITICAL: HTTP/1.1 503 Service Unavailable - string 'Pacman' not found on 'https://bbs.archlinux32.org:443/' - 2636 bytes in 0.034 second response time
[07:39:38] <abaumann> yes, that's me :-)
[07:49:47] <abaumann> https://github.com
[07:49:48] <phrik> Title: GitHub - fluxbb/converter: A tool for converting from other forum software to FluxBB (at github.com)
[07:51:07] <abaumann> is a nice project to merge fluxbb forums, the only problem I have currently is that the old forum was not an official release, so I cherry-picked the new password method of tyzoid (see https://git.archlinux32.org), for now it's more important the new users can still log in, the old users can reset their password if needed..
[07:51:07] <phrik> Title: fluxbb - Archlinux32 customized Fluxbb installation (at git.archlinux32.org)
[07:51:29] <abaumann> ..forums can be joined manually (I will do that) joining old and new posts.
[07:52:27] <abaumann> I also spotted some 5 or 6 accounts with different usernames and same email address, so I'm assuming the new username is ok for future posts. The old accounts remain there, but are inactive (for instance andreas_baumann is inactive, abaumann is active).
[07:53:27] <abaumann> The password mechanism I want to change in a second step (not right now), I have to have a closer look at it. Also I want to make sure we can later upgrade to an official fluxxbb release.
[07:53:43] <abaumann> comments, alternatives, critisism welcome :-)
[07:56:56] <abaumann> mmh. I wonder what happens if I change usernames in the old (or new database) before the merge for the duplicates?
[08:11:29] -!- skydrome has joined #archlinux32
[08:12:14] -!- deep42thought has joined #archlinux32
[08:12:14] <buildmaster> Hi deep42thought!
[08:12:14] <buildmaster> !rq deep42thought
[08:12:15] <phrik> buildmaster: <deep42thought> my supervisor when I made my first shot with the laser: He handed me the remote and said: "Hey Erich, want to destroy a 20 Mio€ device?"
[08:12:38] <deep42thought> abaumann: changing old usernames to match the new ones sound like the correct approach
[08:13:15] <buildmaster> pentium4/wireguard-arch is broken (says nlopc46): https://archlinux32.org
[08:16:47] <buildmaster> pentium4/wireguard-lts are broken (says nlopc46): https://archlinux32.org
[08:17:26] <abaumann> morning deep42thought
[08:17:31] <deep42thought> Hi abaumann!
[08:17:43] <abaumann> yeah but, the old name is sometimes user and the new one user2
[08:18:31] <deep42thought> so the user assumes to be able to log in with "user2", now
[08:18:39] <abaumann> mmh. true that.
[08:18:53] <abaumann> it only affects one person (and me) :-)
[08:24:10] <abaumann> mmh, it only affects me, the others are duplicates with same email with almost identical users ids (numbers), so I suppose those are duplicate registrations.
[08:24:26] <abaumann> and they all have high user ids, so it's after the new forum has been installed.
[08:24:33] <deep42thought> only one forum use, who did not remember his old name?
[08:24:34] <deep42thought> ;-)
[08:24:41] <abaumann> :-)
[08:25:56] <abaumann> mmh. If I change a username, it doesn't affect the forum posts..
[08:26:04] <abaumann> ..sounds like an "interesting" database schema. :-)
[08:26:17] <deep42thought> sounds like too much caching
[08:26:46] <abaumann> select edited_by from posts: is a varchar containing the usersname
[08:27:11] <abaumann> most likely to allow the username to be shown even when the user has been deleted
[08:27:39] <abaumann> select * from * where value_in_table = 'andreas_baumann' :-)
[08:36:10] <abaumann> mmh. this is not a good idea.. I leave both users in there, one with a disabled email..
[08:50:15] <abaumann> mmh. the maintainance mode in fluxbb is.. well.. weird..
[08:56:50] <buildmaster> pentium4/signal-desktop is broken (says nlopc46): https://archlinux32.org
[08:59:23] <trotz> 2020/02/13 08:58 OK buildmaster Forum HTTP OK: HTTP/1.1 200 OK - 15064 bytes in 0.040 second response time
[08:59:24] <trotz> 2020/02/13 08:58 OK buildmaster Forum Secure HTTP OK: HTTP/1.1 200 OK - 14825 bytes in 0.575 second response time
[09:02:04] -!- torv has joined #archlinux32
[09:10:21] <abaumann> mmh. all forums are in double after conversion, ok, well, I have to move all topics from the new forum into the old forum, should not be a problem.
[09:10:52] <deep42thought> the forum runs a raid1, now?
[09:11:01] <abaumann> lol
[09:11:22] <abaumann> no, old posts in old topics, new posts in new topics
[09:11:43] <abaumann> I'll keep the new annoucments of the new forum and the old of the old, right?
[09:12:06] <deep42thought> feel free to merge them
[09:13:23] <trotz> 2020/02/13 09:12 CRIT buildmaster Forum HTTP CRITICAL: HTTP/1.1 503 Service Unavailable - string 'Pacman' not found on 'http://bbs.archlinux32.org:80/' - 2880 bytes in 0.036 second response time
[09:13:23] <trotz> 2020/02/13 09:12 CRIT buildmaster Forum Secure HTTP CRITICAL: HTTP/1.1 503 Service Unavailable - string 'Pacman' not found on 'https://bbs.archlinux32.org:443/' - 2641 bytes in 0.035 second response time
[09:13:41] <abaumann> basically, all admin posts in the new announcment forum are recovered ones, so the old ones have the correct author.
[09:13:52] <deep42thought> ah, right
[09:14:04] <deep42thought> well, then "merge" means "remove the duplicates"
[09:14:10] <abaumann> yep
[09:14:15] <abaumann> in this case
[09:14:21] <abaumann> otherwise the sets should be disjunct
[09:15:20] <deep42thought> and if a post slips your attention and we have duplicates, it's not that tragic at all ...
[09:15:41] <abaumann> I'm more worried about slipping and loosing some posts, actually :-)
[09:15:52] <deep42thought> yes
[09:16:07] <abaumann> ..now come the easy cases.. :-)
[09:16:16] <deep42thought> so rather go the "more likely duplicates, less likely loss" way than the other way round
[09:37:06] <abaumann> so. the merged forum is active.
[09:37:44] <deep42thought> cool, thank you very much!
[09:37:47] <abaumann> np
[09:37:53] <abaumann> was on my todo list for a long time
[09:38:11] <abaumann> try if you can still log in.. :-)
[09:38:46] <deep42thought> Wrong username and/or password.
[09:38:47] <deep42thought> :-)
[09:38:53] <abaumann> ah. common.
[09:39:15] <abaumann> I tested with mine.. so why this?
[09:39:28] <deep42thought> is this my old password, maybe?
[09:39:32] <abaumann> yes.
[09:39:45] <deep42thought> I don't have that anymore :-D
[09:39:53] <abaumann> ah well.
[09:39:57] <abaumann> you can reset the password.
[09:40:00] <deep42thought> will do
[09:40:23] <abaumann> this will also affect some others.. maybe I should make a forum note..
[09:40:42] <deep42thought> write an announcement, yeah
[09:41:13] <abaumann> I also lost the Archlinux32 customization.. *puzzle8
[09:41:15] <deep42thought> yup, new password works
[09:42:04] <abaumann> mmh. maybe the old style is stored in the profiles of the users.
[09:42:16] <abaumann> and the Archlinux32 style was a hacked Archlinux style before
[09:42:39] <abaumann> well. we will have some small glitches, I'm sure..
[09:42:48] <deep42thought> np :-)
[09:43:05] <abaumann> better to have all the content back at their right position (topic id)
[09:43:26] <trotz> 2020/02/13 09:42 OK buildmaster Forum HTTP OK: HTTP/1.1 200 OK - 17120 bytes in 0.040 second response time
[09:43:27] <trotz> 2020/02/13 09:42 OK buildmaster Forum Secure HTTP OK: HTTP/1.1 200 OK - 16881 bytes in 0.053 second response time
[09:45:39] <abaumann> mmh, the Archlinux32 style stubbornly doesn't show up..
[09:46:08] * abaumann closes the browser and reopens it..
[09:47:34] <abaumann> mmh. the links are correct, just the logo not
[09:47:41] <abaumann> ah. that's via config.
[09:59:17] <abaumann> <link rel="stylesheet" type="text/css" href="style/ArchLinux32.css" /> <link rel="stylesheet" media="screen" href="style/ArchLinux/arch.css?v=5"/> <link rel="stylesheet" media="screen" href="style/ArchLinux/archnavbar.css?v=5"/>
[09:59:21] <abaumann> what the duck?
[10:01:01] <abaumann> I really hate this kind of caching.. just make the up without a cache and add varnish on top.
[10:01:04] <abaumann> *app
[10:05:38] <abaumann> aha. got it.
[10:05:45] <abaumann> wrong in Archinux32.css
[10:05:46] <deep42thought> :-)
[10:06:14] <abaumann> we have to have the proper 32-bit logo :-)
[10:06:22] <deep42thought> yes, of course
[10:06:30] <abaumann> "Newest registered user: Panties"
[10:06:34] <abaumann> ok. well. :-)
[10:07:03] <deep42thought> one needs to register his clothing, too?
[10:07:10] <abaumann> of course. :-)
[10:08:51] <abaumann> so, no to more interesting stuff. the sudo-security disaster..
[10:09:33] <deep42thought> the bogus glibc is also in the archive https://archive.archlinux32.org
[10:09:33] <phrik> Title: Index of /packages/g/glibc/ (at archive.archlinux32.org)
[10:09:39] <deep42thought> in case you do not want to rebuild it
[10:09:49] <deep42thought> (my local rebuild yielded the same error)
[10:10:37] <abaumann> ah ok.
[10:14:04] <abaumann> Isn't this sudo bug in the wild for a while?
[10:14:06] -!- T`aZ has quit [Remote host closed the connection]
[10:14:32] <abaumann> CVE-2019-18634, stack overflow issue, of course
[10:14:35] <deep42thought> idk, I just noticed, it also disables ssh logins
[10:14:59] <abaumann> but I don't think glibc fixed the bug in sudo, and now sudo breaks.
[10:15:14] <abaumann> also I think the problems with ssh are completely unrelated to that
[10:15:25] <deep42thought> they are related to glibc
[10:15:30] <deep42thought> downgrading glibc helped
[10:15:43] <abaumann> and 64-bit upstream had no issues?
[10:16:09] <deep42thought> none that I'm aware of :-/
[10:16:27] <deep42thought> it probably boils down to "we need to build package $x before updating glibc"
[10:16:47] <deep42thought> or "we need to build intermediate versions of glibc,$x,... before building their final versions"
[10:18:41] <abaumann> ok, the sudo bug is just a stack overflow when printing the password feedback onto the screen, which is a completely bogus feature anyway.
[10:18:53] <abaumann> pwfeedback
[10:34:11] <buildmaster> pentium4/python-shapely is broken (says nlopc46): https://archlinux32.org
[10:43:10] <abaumann> afk, shoppine before the rain comes..
[10:43:16] <abaumann> *shoppinge
[10:43:20] <deep42thought> cu
[10:43:24] <abaumann> cu
[11:10:53] -!- T`aZ has joined #archlinux32
[11:30:19] * buildmaster failed to execute a mysql query - can you have a look at "tmp.mysql-functions.query.2020-02-13T11:28:06.z4KvC7.stdin"?.
[11:33:14] * buildmaster resumes sanity.
[11:35:55] * buildmaster failed to execute a mysql query - can you have a look at "tmp.mysql-functions.query.2020-02-13T11:33:31.4jfFac.stdin"?.
[11:36:09] <deep42thought> damn those typos
[11:37:55] * buildmaster resumes sanity.
[11:39:24] <deep42thought> hmm, strange, I cannot reproduce the ssh login problem with the new glibc-2.31-1.0 on pentium4
[11:39:33] -!- bill-auger_ has quit [Read error: Connection reset by peer]
[11:40:15] -!- bill-auger has joined #archlinux32
[11:46:16] <deep42thought> but the sudo error persists
[11:47:00] <deep42thought> (maybe it's not a sudo error, but rather an error of something called via sudo)
[12:06:03] -!- DepositePirate has quit [Ping timeout: 240 seconds]
[12:08:13] -!- DepositePirate has joined #archlinux32
[12:21:08] <buildmaster> pentium4/virtualbox-modules-arch is broken (says nlopc46): https://archlinux32.org
[12:25:27] <buildmaster> pentium4/haskell-network-uri is broken (says nlopc46): https://archlinux32.org
[12:29:32] <buildmaster> pentium4/haskell-http-client is broken (says nlopc46): https://archlinux32.org
[12:35:03] <buildmaster> pentium4/haskell-http is broken (says nlopc46): https://archlinux32.org
[12:36:20] <buildmaster> pentium4/haskell-warp is broken (says eurobuild6-1): https://archlinux32.org
[12:37:50] <abaumann> mmh. I just reinstalled slave1-6 and it builds fine.
[12:38:03] <abaumann> so you think problems would start as soon we use new glibc?
[12:38:13] <abaumann> ok, that's something to test. :-)
[12:39:30] <buildmaster> pentium4/haskell-criterion is broken (says nlopc46): https://archlinux32.org
[12:41:47] <abaumann> (2/4) Creating temporary files...
[12:41:47] <abaumann> [/usr/lib/tmpfiles.d/systemd-tmp.conf:11] Failed to replace specifiers: /tmp/systemd-private-%b-*
[12:41:50] <abaumann> [/usr/lib/tmpfiles.d/systemd-tmp.conf:12] Failed to replace specifiers: /tmp/systemd-private-%b-*/tmp
[12:41:53] <abaumann> [/usr/lib/tmpfiles.d/systemd-tmp.conf:13] Failed to replace specifiers: /var/tmp/systemd-private-%b-*
[12:41:56] <abaumann> [/usr/lib/tmpfiles.d/systemd-tmp.conf:14] Failed to replace specifiers: /var/tmp/systemd-private-%b-*/tmp
[12:41:59] <abaumann> [/usr/lib/tmpfiles.d/systemd-tmp.conf:22] Failed to replace specifiers: /var/lib/systemd/coredump/.#core*.%b*
[12:42:02] <abaumann> fchmod() of /run/systemd/netif failed: No such file or directory
[12:42:05] <abaumann> fchmod() of /run/systemd/netif/links failed: No such file or directory
[12:42:07] <abaumann> fchmod() of /run/systemd/netif/leases failed: No such file or directory
[12:42:09] <abaumann> ACL operation on "/var/log/journal" failed: No such file or directory
[12:42:12] <abaumann> ACL operation on "/var/log/journal" failed: No such file or directory
[12:42:14] <abaumann> Failed to re-open '/var/log/journal': No such file or directory
[12:42:17] <abaumann> Failed to re-open '/var/log/journal/remote': No such file or directory
[12:42:20] <abaumann> error: command failed to execute correctly
[12:42:22] <abaumann> When installing glibc into a chroot. This makes extremely happy if glibc depends now on systemd. :->
[12:42:25] <abaumann> &me
[12:42:28] <abaumann> *This makes me happy
[12:43:19] <abaumann> tons of 'warning: warning given when extracting /usr/share/i18n/locales/nhn_MX (Can't restore time)
[12:43:22] <abaumann> warning: warning given when extracting /usr/share/i18n/locales/niu_NU (Can't restore time)
[12:43:25] <abaumann> warning: warning given when extracting /usr/share/i18n/locales/niu_NZ (Can't restore time)
[12:43:54] <abaumann> but, I could build a which with it (not one of the flying ones with a t)
[13:17:43] <abaumann> mmh. archlinux32.org has quite some mysql load..
[13:30:39] <abaumann> I restarted mysql and nginx on archlinux32.org
[13:30:46] <deep42thought> re
[13:32:47] <deep42thought> you say, you're able to build "which" with the new glibc?
[13:32:52] <deep42thought> this sounds interesting
[13:33:28] <abaumann> no, I'm building with the old one.
[13:33:34] <abaumann> to kill the backlog
[13:33:40] <deep42thought> ok
[13:33:45] <abaumann> why is start slave on archlinux32.org not starting the mysql slave?
[13:34:51] <deep42thought> because it has thrown some error
[13:35:03] <deep42thought> Error 'Table 'buildmaster.rd_q' doesn't exist' on query.
[13:35:26] <abaumann> where is this error message?
[13:35:33] <deep42thought> show slave status
[13:35:59] <deep42thought> in root's `mysql` prompt
[13:38:14] <deep42thought> should I reclone or are you up for the hunt of the true error?
[13:38:29] <abaumann> nah. reclone it.
[13:38:33] <deep42thought> ok
[13:38:50] <abaumann> I guess it's the usual bug: transactions are not transactions on mysql
[13:38:57] <deep42thought> !grab abaumann
[13:38:57] <phrik> deep42thought: Tada!
[13:42:59] <deep42thought> hehe: https://git.archlinux32.org
[13:42:59] <phrik> Title: re-bootstrap-slave-db - devops - Archlinux32 devs' convenience-scripts (at git.archlinux32.org)
[13:43:21] <deep42thought> a.k.a. "I'm not doing this for the first time"
[13:43:33] <abaumann> :-)
[14:01:01] <deep42thought> man, this takes ages ...
[14:06:36] <abaumann> uh. sorry for breaking it..
[14:06:49] <deep42thought> I think, it was not your fault
[14:06:59] <abaumann> how rare ;-)
[14:07:10] <deep42thought> lol
[14:10:25] <deep42thought> oh, it finished :-)
[14:12:02] <abaumann> cool :-)
[14:15:12] <buildmaster> i686/octave is broken (says eurobuild6-6): https://archlinux32.org
[14:20:56] <deep42thought> btw: don't worry about the last week(s) showing no green dots: it was a logging-glitch only: return-assignment never returned with exit-code 0 in case of a successfully built package, but the package was nevertheless successfully uploaded
[14:21:10] <abaumann> ah. that's good news.
[14:21:55] <deep42thought> it just made the build slaves upload every package *twice* to the buildmaster: the first being uploaded to the mirror and the second being rejected as "you don't build that anymore"
[14:22:29] <abaumann> this helps against ssh glitches?
[14:22:53] <deep42thought> or any other temporary failure, yes
[14:23:11] <deep42thought> e.g. "deep42thought reboots the buildmaster without warning the slaves"-errors
[14:23:29] <abaumann> ah :-)
[14:23:46] <deep42thought> also more importantly against lock congestion
[14:26:43] <abaumann> So, another one for the gallery: https://bbs.archlinux32.org
[14:26:45] <phrik> Title: IBM Thinkpad 240X / Artwork, Screenshots & Setups / Arch Linux 32 Forum (at bbs.archlinux32.org)
[14:29:33] <deep42thought> this is i686, right?
[14:29:36] <abaumann> yes.
[14:29:48] <deep42thought> just asking, because you wrote, you booted from i486
[14:29:48] <abaumann> i486 boot ISO
[14:29:50] <abaumann> oups.
[14:29:54] <deep42thought> ok :-)
[14:29:57] <abaumann> good catch. :-)
[14:31:02] <deep42thought> nit-picker should have said that ...
[14:31:45] <buildmaster> i686/gmic is broken (says eurobuild6-5): https://archlinux32.org
[14:33:54] <buildmaster> i686/flashplugin is broken (says nlopc46): https://archlinux32.org
[14:37:28] <buildmaster> i686/mame is broken (says eurobuild6-2): https://archlinux32.org
[14:38:22] <buildmaster> i686/purpose is broken (says eurobuild6-3): https://archlinux32.org
[15:24:58] <nit-picker> key A0B250C0FC9FC079EC04ADB7A50C0F20AEC3AF00 (from Polichronucci (Arch Linux 32 Master Key) <polichronucci@archlinux.gr>) in package archlinux32-keyring-transition-20200113-1.0-any.pkg.tar.zst expires on 2020-05-05 (in 81 < 100 days).
[15:25:29] <buildmaster> any/python-os-client-config is broken (says eurobuild6-2): https://archlinux32.org
[15:37:10] -!- deep42thought has quit [Quit: Leaving.]
[16:10:22] <nit-picker> key A0B250C0FC9FC079EC04ADB7A50C0F20AEC3AF00 (from Polichronucci (Arch Linux 32 Master Key) <polichronucci@archlinux.gr>) in package archlinux32-keyring-20200113-1.0-any.pkg.tar.zst expires on 2020-05-05 (in 81 < 100 days).
[16:12:26] -!- abaumann has quit [Quit: leaving]
[17:09:48] <trotz> 2020/02/13 17:09 CRIT buildmaster OS updates 1 updates, 0 ignored
[17:14:14] -!- abaumann has joined #archlinux32
[17:14:14] <buildmaster> Hi abaumann!
[17:14:14] <buildmaster> !rq abaumann
[17:14:15] <phrik> buildmaster: <abaumann> All the times I was debugging some network settings, just to find out, that I tripped a cable. :-)
[17:14:19] -!- deep42thought has joined #archlinux32
[17:14:19] <buildmaster> Hi deep42thought!
[17:14:19] <buildmaster> !rq deep42thought
[17:14:20] <phrik> buildmaster: <deep42thought> systemd-naked?
[17:14:24] <deep42thought> Hi abaumann!
[17:14:26] <abaumann> hi deep42thought
[17:14:32] <abaumann> wow, that was fast. :-)
[17:14:36] <abaumann> was that you or a robot?
[17:14:42] <abaumann> sudo: error in event loop: Operation not permitted
[17:14:42] <abaumann> sudo: unexpected child termination condition: 0
[17:14:44] <deep42thought> it was me
[17:14:52] <abaumann> I gould reproduce it with a self-compiled new glibc too.
[17:14:54] <abaumann> *could
[17:15:03] <abaumann> so, now for debugging the thing..
[17:15:23] <deep42thought> good hunting!
[17:16:28] <abaumann> yeah, thanks. :-)
[17:57:51] <abaumann> wow. makechrootpkg writes a /chrootbuild script into the chroot by printing the _chrootbuild function..
[17:58:16] <abaumann> + sudo --preserve-env=SOURCE_DATE_EPOCH -iu builduser bash -c 'cd /startdir; makepkg "$@"' -bash --syncdeps --noconfirm --log --holdver --skipinteg --install
[17:58:19] <abaumann> sudo: error in event loop: Operation not permitted
[17:58:21] <abaumann> sudo: unexpected child termination condition: 0
[17:59:35] <abaumann> more /var/lib/archbuild/staging-pentium4/abaumann/etc/sudoers.d/builduser-pacman
[17:59:38] <abaumann> builduser ALL = NOPASSWD: /usr/bin/pacman
[17:59:52] <abaumann> so, yes, the build user is allowed to run pacman, but not to run makepkg
[18:00:09] <abaumann> don't tell me sudo was so broken, that it actually worked by accident before..
[18:01:10] <abaumann> let's see what the upstream chroot is doing..
[18:01:51] <trotz> 2020/02/13 18:00 CRIT okeanos Home Certificate HTTP CRITICAL - Invalid HTTP response received from host on port 443: cURL returned 28 - Timeout was reached
[18:03:13] <abaumann> so staging-x86_64 runs with a glibc 2.31-1, mmh.
[18:04:52] <trotz> 2020/02/13 18:03 CRIT okeanos Home Secure HTTP CRITICAL - Invalid HTTP response received from host on port 443: cURL returned 28 - Timeout was reached
[18:07:52] <trotz> 2020/02/13 18:07 CRIT okeanos Home HTTP CRITICAL - Invalid HTTP response received from host on port 80: cURL returned 28 - Timeout was reached
[18:11:52] <trotz> 2020/02/13 18:11 OK okeanos Home Certificate OK - Certificate 'archlinux32.org' will expire on Sat 02 May 2020 12:10:31 AM GMT +0000.
[18:13:52] <trotz> 2020/02/13 18:13 OK okeanos Home Secure HTTP OK: HTTP/2.0 200 - 23361 bytes in 1.400 second response time
[18:16:09] <buildmaster> any/gunicorn is broken (says eurobuild6-1): https://archlinux32.org
[18:16:10] <buildmaster> any/python-paste is broken (says eurobuild6-4): https://archlinux32.org
[18:17:52] <trotz> 2020/02/13 18:17 OK okeanos Home HTTP OK: HTTP/1.1 200 OK - 23273 bytes in 1.193 second response time
[18:18:17] -!- T`aZ has quit [Remote host closed the connection]
[18:23:48] <deep42thought> before complaining too furiously about upstream (either of both), keep in mind, that we forked devtools, so it might also be a bug on our end :-/ though: why should it work with the old glibc, then
[18:43:37] <abaumann> I'm actually not complaining.. I'm merely comparing.. to find differences.
[18:44:04] <abaumann> though complaining about upstream is something (I admit) I'm usually doing quite fast. :-)
[18:44:05] <deep42thought> ok, then good, carry on :-)
[18:44:31] <abaumann> so far I have no clue, I have to debug a sudo inside the chroot
[18:45:02] <abaumann> one suspicion (which always pops up) is some security thingy not working for IA-32
[18:45:10] <abaumann> libsecomp for instance
[19:28:27] -!- skydrome has quit [Quit: Birds are and always have been reincarnated old men with Tourette's syndrome having somehow managed to dupe the reproductive saga.]
[19:29:30] <abaumann> it's an internal event loop listening to the internal sudo to called command communication which fails with operation not permited
[19:30:02] <abaumann> touch: setting times of '/usr': Operation not permitted
[19:30:02] <abaumann> error: command failed to execute correctly
[19:30:54] <abaumann> there is another possibility: some glibc hooks are broken and make sudo fail when it tries to install packages via pacman
[19:33:05] <abaumann> the operation permited doesn't come from sudo, but from the process executed further down _by_ sudo
[19:33:26] * abaumann picks at the hooks in glibc
[19:33:45] -!- skydrome has joined #archlinux32
[19:34:09] <deep42thought> can you try su instead of sudo?
[19:34:19] <abaumann> yes, I can.
[19:34:30] <abaumann> Actually it could be any hook using some functionality of the chroot
[19:36:27] <abaumann> bash -c 'cd /startdir; makepkg "$@"' -bash "$@"
[19:36:28] <abaumann> *sigh*
[19:36:57] <abaumann> actually, is the chroot spawned as root and then drops privileges to builduser?
[19:37:00] <abaumann> with sudo?
[19:37:10] <deep42thought> yes, I think, it's that way round
[19:37:18] <deep42thought> but I could be wrong here
[19:37:33] * abaumann boldly executes things as root inside the chroot
[19:37:59] <abaumann> ==> ERROR: Running makepkg as root is not allowed as it can cause permanent, catastrophic damage to your system.
[19:38:02] <abaumann> yep :-)
[19:40:42] <abaumann> sorry, I can not rewrite this..this is to complicated for my head.
[19:41:04] <deep42thought> rewrite?
[19:41:36] <abaumann> mmh. checking out a fresh version of devtools..
[19:42:07] <abaumann> sudo --preserve-env=SOURCE_DATE_EPOCH -iu builduser bash -> su?
[19:45:10] <buildmaster> i686/python-shapely is broken (says eurobuild6-6): https://archlinux32.org
[19:45:24] <abaumann> default:
[19:45:24] <abaumann> sudo_warnx(U_("unexpected child termination condition: %d"), cstat.type);
[19:45:34] <abaumann> yeah: we get into an unusal situation in sudo.
[19:45:37] <abaumann> but why.
[19:46:07] <abaumann> I'm worried about all those timestamp warning when installing a new glibc
[19:46:18] <abaumann> I think, that's the root cause
[19:46:47] <deep42thought> yeah, that's definitely troublesome
[19:47:13] <abaumann> usually glibc should not install with tons of warnings.
[19:47:23] <abaumann> the systemd ones are not dramatic, but the timestamps are.
[19:47:50] <abaumann> when I can trigger a normal command in the chroot also exhibting that behaviour, for instance touch, stat or so..
[19:48:01] <abaumann> ..then it's easier to narrow down the bug.
[20:06:16] -!- isacdaavid has joined #archlinux32
[20:06:56] <buildmaster> any/python-aioresponses are broken (says eurobuild6-4): https://archlinux32.org
[20:15:41] <abaumann> mmh. the timestamps might be a ntp problem.
[20:15:50] <abaumann> or more not intalling one. :-)
[20:15:52] <abaumann> *installing
[20:16:09] <abaumann> but the sudo operation not permitted problem persists
[20:16:33] <abaumann> the interesting thing is: if I call 'strace -f sudo' instead of just sudo, the building in the slave starts.
[20:16:42] <abaumann> this sounds like a race-condition to me
[20:18:23] -!- trotz has quit [Remote host closed the connection]
[20:18:24] <deep42thought> workaround-fix: we call sudo via strace
[20:18:30] <deep42thought> either it works and we're done
[20:18:38] <deep42thought> or it fails and we can look into the issue :-D
[20:18:55] <abaumann> nah. :-)
[20:19:02] <abaumann> it's not really deterministic.
[20:19:46] <deep42thought> anyways: if it fails, you get an strace :-)
[20:19:51] -!- buildmaster has quit [Remote host closed the connection]
[20:20:07] <abaumann> upgrading systemd and then getting network is unreachable..
[20:20:15] <abaumann> ..well, this I had before..
[20:20:34] <abaumann> now ssh hangs while connecting
[20:20:45] <abaumann> yeah. shutdown issues.
[20:20:57] <abaumann> I just updated the buildmaster..
[20:21:00] -!- trotz has joined #archlinux32
[20:21:01] <trotz> 2020/02/13 17:09 CRIT buildmaster OS updates 1 updates, 0 ignored
[20:21:13] <deep42thought> this are issues on the buildmaster?
[20:21:20] <abaumann> updates
[20:21:21] <abaumann> :-)
[20:21:29] <deep42thought> ah, ok
[20:21:34] <abaumann> systemd updates.. no issues (yet) :-)
[20:21:35] <deep42thought> I thought the ssh issue was on the buildmaster
[20:21:39] <deep42thought> :-D
[20:21:40] <abaumann> nono
[20:21:43] -!- buildmaster has joined #archlinux32
[20:21:43] <buildmaster> !rq buildmaster
[20:21:44] <phrik> buildmaster: <buildmaster> I might be insane, but never confused ... ;-)
[20:21:46] <abaumann> all good.
[20:23:18] <deep42thought> gnaaaaa
[20:23:19] <deep42thought> stupid me
[20:23:32] <deep42thought> I assumed, all build slaves could get information from the database
[20:23:39] <deep42thought> but of course, they cannot
[20:24:05] * deep42thought would bang his head on the table, but he has an alix infront of his keyboard
[20:26:56] <abaumann> huh?
[20:27:03] <abaumann> they died now?
[20:27:06] <deep42thought> yes
[20:27:28] <abaumann> ERROR: Unknown exit code 255 from 'get-assignment'.
[20:27:29] <abaumann> just one
[20:27:44] <deep42thought> no, sending the packages back does not work
[20:27:50] <abaumann> aeh. correction. just all. :-)
[20:27:51] <deep42thought> e.g. no packages are sent at all
[20:27:59] <deep42thought> 'Your buildslave "rechenknecht" uploaded a package with a not fully-trusted signature:'
[20:28:04] <deep42thought> without any additional info
[20:28:55] <abaumann> ..and again I'm breaking things.. also called testing.. :-)
[20:32:39] <abaumann> sleep 5
[20:32:39] <abaumann> sleep: cannot read realtime clock: Operation not permitted
[20:32:41] <abaumann> interesting..
[20:32:55] * deep42thought has seen this, too
[20:33:15] <abaumann> but not when I use the chroot with arch-chroot, only during makechrootpkg
[20:33:23] <abaumann> which is started by systemd-spawn
[20:33:45] <deep42thought> I put a `sleep 60` into prepare() to be able to extract glibc manually during build, but it failed with your error
[20:34:01] <abaumann> and I put a sleep 5 before sudo to see if it works then :-)
[20:37:56] <abaumann> so, glibc works with privileges, so that systemd-nspawn triggers security issues..
[20:38:13] <deep42thought> is systemd up-to-date?
[20:38:20] <deep42thought> should we rebuild that, too, maybe?
[20:38:26] <abaumann> rebuild? how?
[20:38:36] <deep42thought> I mean before glibc
[20:38:42] <abaumann> and isn't that the systemd-nspan from 64-bit (from the host)?
[20:38:49] <abaumann> so rebuilding it in the chroot doesn't help
[20:38:50] <deep42thought> oh, right
[20:39:59] <abaumann> maybe nspawn can be convinced, not to drop privileges
[20:40:08] <abaumann> because this might need some glibc support..
[20:40:15] <abaumann> ..but here I'm really on thin ice..
[20:49:13] <abaumann> ==> Appending keys from archlinux32.gpg...
[20:49:13] <abaumann> gpg: can't connect to the agent: IPC connect call failed
[20:49:15] <abaumann> + pacman-key --populate
[20:49:27] <abaumann> I'm trying to set capabilities by hand now.
[20:53:50] <abaumann> mmh. system call filters via libsecomp eventually?
[20:56:07] <abaumann> mmh. if libseccomp filters would fail, I would expect the famous ones as in libmagic/file, but not RTC and time..
[20:56:43] -!- DepositePirate has quit [Remote host closed the connection]
[20:56:57] <abaumann> but assuming the host-systemd sets some seccomp-filters and the new glibc maybe calls them differently or not, this would explain the troubles we se.
[20:57:00] <abaumann> *see
[20:57:25] -!- DepositePirate has joined #archlinux32
[20:57:43] <abaumann> but who would add libseccomp filtering to glibc?
[20:58:04] <deep42thought> what does libseccomp do?
[20:58:34] <abaumann> it basically makes sure you don't call a kernel syscall, unless you declared somewhere before, you want to use it.
[20:58:46] <deep42thought> hmm
[20:59:15] <abaumann> basically you can say: I use 'open' to read libmagic, but then I'm not opening any files anymore while reading the file to be analyzed via a buffer.
[20:59:43] <deep42thought> so you can drop capabilities, right?
[20:59:49] <abaumann> and as syscalls are not the same between 32-bit and 64-bit in Linux and people usually only adapt the later..
[21:00:02] <deep42thought> ah, I see
[21:00:27] <deep42thought> "I don't call any 32-bit commands"; "setarch i686"
[21:00:38] <abaumann> mmh? where?
[21:00:46] <deep42thought> no, just a thought
[21:01:04] <abaumann> ah.
[21:01:19] <deep42thought> sry, I'm currently one-handed ...
[21:01:29] <deep42thought> soooo slow
[21:01:30] <abaumann> the build system sometimes calls arch-nspawn, sometimes systemd-nspawn directly.
[21:01:41] <abaumann> oh. bathing children? ;-)
[21:01:46] <deep42thought> when does it do the latter?
[21:02:29] <abaumann> nah. actually. it should always go over the wrapper, so I can do some experiments there.
[21:02:52] <deep42thought> :-)
[21:03:23] <deep42thought> not bathing, just swaying
[21:04:01] <abaumann> :-)
[21:04:55] <abaumann> systemd-analyze syscall-filter actually knows about 32-bit seccomp filter names
[21:04:56] <deep42thought> they're so peaceful, when they sleep ;-P
[21:05:08] <abaumann> :-)
[21:05:12] <deep42thought> ok, and it translates correctly, hopefully
[21:05:13] <abaumann> when, that is.
[21:07:18] <abaumann> systemd-analyze syscall-filter aha
[21:07:28] <abaumann> they have groups, like @clock
[21:07:46] <abaumann> and at the end I see a list of ungrouped sycalls, like timer_gettime32
[21:08:13] <abaumann> now: try to solve the "error message" to "syscall" puzzle :-)
[21:10:53] <deep42thought> use a raibow-table?
[21:11:04] <abaumann> :-)
[21:22:02] <abaumann> clock_nanosleep_time64(CLOCK_REALTIME, 0, {tv_sec=5, tv_nsec=0}, 0xffa8184c) = -1 EPERM (Operation not permitted)
[21:22:13] <abaumann> when I strace sleep 5
[21:23:22] <abaumann> --system-call-filter="clock_nanosleep_time64" doesn't have an effect, why?
[21:24:42] <abaumann> another problem is when running a 32-chroot on a 64-bit-host, this might play funny games with syscall names
[21:25:44] <abaumann> well.. tomorrow.. :-)
[21:25:49] <deep42thought> good night :-)
[21:25:56] <abaumann> cu :-)
[21:25:57] -!- abaumann has quit [Quit: leaving]
[21:34:36] -!- renf has joined #archlinux32
[21:34:39] <renf> Hello!
[21:34:49] <deep42thought> Hi renf
[21:35:06] <renf> What is the minimum amount of RAM I need for 32 bit Arch?
[21:35:34] <deep42thought> I think, the absolute minimum is 64MB for the i486 branch
[21:35:37] <deep42thought> iirc
[21:36:02] <renf> I have a system with 64MB and i686...
[21:36:21] <renf> Is there a way to check this out?
[21:36:34] <deep42thought> a live disk?
[21:36:39] <deep42thought> https://bbs.archlinux32.org
[21:36:41] <phrik> Title: i486 motivation thread / Artwork, Screenshots & Setups / Arch Linux 32 Forum (at bbs.archlinux32.org)
[21:38:09] <deep42thought> I think, the normal initramdisk will not work there, hmm
[21:38:21] <deep42thought> pxe boot might work
[21:39:14] <renf> What's pxe boot?
[21:39:32] <deep42thought> booting via network
[21:39:59] <deep42thought> https://wiki.archlinux.org
[21:39:59] <phrik> Title: PXE - ArchWiki (at wiki.archlinux.org)
[21:40:06] -!- abaumann has joined #archlinux32
[21:40:06] <buildmaster> Hi abaumann!
[21:40:06] <buildmaster> !rq abaumann
[21:40:07] <phrik> buildmaster: <abaumann> I consider switching to Outlook and Windows ;-)
[21:40:12] <deep42thought> ah, here comes the expert :-)
[21:40:18] <abaumann> Actually, 64-bit time functions in 32-bit on 64-bit host, selinux and seccomp come to my mind. Glibc has 64-bit time functions now? Or the kernel?
[21:40:40] <deep42thought> the first was my thought, too
[21:40:53] <abaumann> Just a last minute intuition. :-)
[21:40:59] <deep42thought> though, I fail to see, why this should create problems
[21:41:19] <deep42thought> the best ideas are right before falling asleep ;-)
[21:41:41] <abaumann> Sadly, yes.. and then you forget them.
[21:42:27] <renf> Wait, why PXE? Does it use less RAM than the normal ramdisk?
[21:42:58] <deep42thought> hmm, sounds wrong
[21:43:49] <deep42thought> abaumann: what's the minimum ram that you had archlinux32 running on?
[21:44:17] <abaumann> 64mb. But with an older kernel
[21:44:35] <abaumann> My pxe i486 boots an nbd device
[21:44:46] <abaumann> No copy to ram. No iso.
[21:44:53] <deep42thought> without initramdisk, right?
[21:45:03] <renf> abaumann: We're trying to figure if my 64MB i686 device can work it
[21:45:17] <abaumann> With a reduced one, i think
[21:45:30] <abaumann> Ah :-)
[21:45:52] <abaumann> The iommu disable trick might be important
[21:46:21] <abaumann> See in the forum. thinkpad 240x
[21:46:52] <abaumann> Now that the forum is fully back.. :-)
[21:47:00] <deep42thought> \o/
[21:47:02] <deep42thought> :-D
[21:48:25] -!- T`aZ has joined #archlinux32
[21:49:14] * abaumann yawns
[21:49:34] -!- abaumann has quit [Quit: Quit]
[21:49:34] <deep42thought> should I sing you a lullaby, too?
[21:49:37] <deep42thought> too late
[21:58:13] -!- renf has quit [Quit: leaving]
[21:59:19] -!- T`aZ_ has joined #archlinux32
[22:02:05] <trotz> 2020/02/13 22:01 CRIT okeanos Home HTTP CRITICAL - Invalid HTTP response received from host on port 80: cURL returned 28 - Timeout was reached
[22:04:42] -!- T`aZ_ has quit [Quit: leaving]
[22:06:05] <trotz> 2020/02/13 22:05 CRIT okeanos Home Certificate HTTP CRITICAL - Invalid HTTP response received from host on port 443: cURL returned 28 - Timeout was reached
[22:16:04] <trotz> 2020/02/13 22:15 OK okeanos Home Certificate OK - Certificate 'archlinux32.org' will expire on Sat 02 May 2020 12:10:31 AM GMT +0000.
[22:20:03] -!- KeiraT has quit [Remote host closed the connection]
[22:22:19] -!- KeiraT has joined #archlinux32
[22:33:36] -!- isacdaavid has quit [Ping timeout: 268 seconds]
[22:38:04] <trotz> 2020/02/13 22:37 CRIT okeanos Home Secure HTTP CRITICAL - Invalid HTTP response received from host on port 443: cURL returned 28 - Timeout was reached
[22:39:08] -!- isacdaavid has joined #archlinux32
[22:40:04] <trotz> 2020/02/13 22:39 CRIT okeanos Home Certificate HTTP CRITICAL - Invalid HTTP response received from host on port 443: cURL returned 28 - Timeout was reached
[22:48:04] <trotz> 2020/02/13 22:47 OK okeanos Home Secure HTTP OK: HTTP/2.0 200 - 23361 bytes in 6.849 second response time
[22:52:04] <trotz> 2020/02/13 22:51 OK okeanos Home HTTP OK: HTTP/1.1 200 OK - 23273 bytes in 1.082 second response time
[22:55:13] -!- deep42thought has quit [Quit: Leaving.]
[23:01:05] <trotz> 2020/02/13 23:00 OK okeanos Home Certificate OK - Certificate 'archlinux32.org' will expire on Sat 02 May 2020 12:10:31 AM GMT +0000.
[23:12:06] <trotz> 2020/02/13 23:11 CRIT okeanos Home Secure HTTP CRITICAL - Invalid HTTP response received from host on port 443: cURL returned 28 - Timeout was reached
[23:22:06] <trotz> 2020/02/13 23:21 OK okeanos Home Secure HTTP OK: HTTP/2.0 200 - 23361 bytes in 1.360 second response time