#archlinux32 | Logs for 2023-04-04

Back
[01:53:09] -!- iamawacko has joined #archlinux32
[05:16:27] -!- epony has joined #archlinux32
[06:13:25] -!- drathir_tor has quit [Remote host closed the connection]
[06:13:48] -!- drathir_tor has joined #archlinux32
[06:28:24] -!- iamawacko has quit [Ping timeout: 255 seconds]
[07:18:09] -!- zxrom_ has joined #archlinux32
[07:19:28] -!- zxrom has quit [Ping timeout: 248 seconds]
[07:19:57] -!- pants has quit [Quit: new kernel]
[07:20:03] zxrom_ is now known as zxrom
[07:25:48] -!- pants has joined #archlinux32
[09:14:09] -!- zxrom_ has joined #archlinux32
[09:15:31] -!- zxrom has quit [Ping timeout: 240 seconds]
[09:27:35] zxrom_ is now known as zxrom
[11:04:05] -!- drathir_tor has quit [Remote host closed the connection]
[11:09:20] -!- drathir_tor has joined #archlinux32
[13:29:39] -!- titus_livius has joined #archlinux32
[13:30:56] -!- buildmaster has joined #archlinux32
[13:44:12] -!- abaumann has joined #archlinux32
[13:44:13] <buildmaster> Hi abaumann!
[13:44:13] <buildmaster> !rq abaumann
[13:44:14] <phrik> buildmaster: <abaumann> *abaumann thinks it's not a good idea to cook and fix bugs in ffmpeg in parallel :-)
[13:44:35] <abaumann> sunshavi: you cannot skip the installation of a file belonging to a package
[13:44:56] <abaumann> you can overwrite it, but that should usually not be necessary and is considered a packaging bug
[13:45:21] <abaumann> deep42thought: well, soon the backup is over and then I'll reenable the build machines.. we'll see what happens.. :-)
[14:17:54] -!- abaumann has quit [Quit: leaving]
[14:55:57] -!- epony has quit [Ping timeout: 268 seconds]
[15:00:49] -!- epony has joined #archlinux32
[15:09:11] <girls> one *can* skip files with pacman
[15:09:28] <girls> "NoExtract = file" in /etc/pacman.conf
[15:09:55] <girls> abaumann: ah, ok, a backup might explain, why everything was so slow :)
[16:55:30] -!- abaumann has joined #archlinux32
[16:55:31] <buildmaster> Hi abaumann!
[16:55:31] <buildmaster> !rq abaumann
[16:55:31] <phrik> buildmaster: <abaumann> I like hacking tools :-)
[16:55:41] <abaumann> mmh, still getting tons of haskell in get-package-updates..
[16:58:30] <abaumann> I put them to blacklist/pentium4/ghc, maybe that is not the right place?
[17:00:37] <girls> let me check
[17:02:06] <girls> looks good in the repo
[17:02:51] <girls> umm, mysql_generate_package_metadate exits unclean for a reason, if it could not create metadata
[17:03:33] <girls> there's -f for get-package-updates to skip these errors
[17:03:44] <girls> I'll revert your hack, now ;)
[17:05:57] <girls> actually, there are only *very* few haskell-* packages on the build-list, now - what exactly is the behaviour, you're seeing and you think is wrong?
[17:07:36] <girls> I'm not 100% sure, but I think, get-package-updates will still track the version of deletion-list packages - i.e. they will still show up in the database and everything, but they end up on the deletion-list, not on the build-list
[17:08:20] <girls> it's only at the stage of applying the patches / combining the PKGBUILDs, that the buildmaster does notice, that a package is on the deletion-list
[17:26:29] <abaumann> a, yes please: my hack is just a hack ;-)
[17:26:42] <girls> :D
[17:26:49] <abaumann> aha.
[17:27:00] <abaumann> so, I see the merging git repos message from get-package-updates..
[17:27:12] <abaumann> ..yeah, actually, no haskell package got built anymore..
[17:27:30] <girls> \o/
[17:40:37] <abaumann> mysql>2 ERROR 1213 (40001) at line 8: Deadlock found when trying to get lock; try restarting transaction
[17:40:40] <abaumann> mysql>2 ERROR 1213 (40001) at line 8: Deadlock found when trying to get lock; try restarting transaction
[17:40:52] <abaumann> yes, I basically cannot trigger anything and force build on the buildmaster arnymore..
[17:46:48] <abaumann> ..so I have to put go15 into git and wait till it gets built (which can take days).
[17:47:05] <abaumann> but the backup is over. :-)
[17:47:35] <abaumann> DELETE `binary_packages_in_repositories` FROM `repositories` JOIN `binary_packages_in_repositories
[17:47:58] <abaumann> yeah, deltes in databases can be a pain.. they where simply not invented for that..
[17:48:05] <abaumann> *deletes
[17:49:24] <abaumann> so, enough for today.. :-) bye
[17:49:26] -!- abaumann has quit [Quit: leaving]
[18:44:41] <girls> hmm, I believe, the excessive blacklisting just put too much pressure on the db - I didn't have in mind blacklisting thousands of packages :)
[18:49:10] <girls> the problem is, that the blacklisted packages will have to be handled on each get-package-updates invocation
[18:49:40] <girls> and if there are 5k packages blacklisted, it effectively behaves as if there were 5k package updates each time (in terms of db operations)
[18:50:32] <girls> because the buildmaster will effectively, after fetching package updates, remove all blacklisted packages from that list via a giant join
[20:12:55] -!- iamawacko has joined #archlinux32
[21:33:06] -!- heartburn has quit [Ping timeout: 250 seconds]
[21:40:24] -!- heartburn has joined #archlinux32