#archlinux32 | Logs for 2023-07-09

Back
[00:18:55] -!- girls has quit [Server closed connection]
[00:21:49] -!- girls has joined #archlinux32
[00:49:35] -!- T`aZ has quit [Server closed connection]
[00:52:51] -!- T`aZ has joined #archlinux32
[02:34:17] -!- finsternis has quit [Read error: Connection reset by peer]
[03:18:13] <KitsuWhooa> ModuleNotFoundError: No module named 'docutils'
[03:18:56] <KitsuWhooa> some packages are failing to build because of that
[03:19:15] -!- bill-auger has quit [Server closed connection]
[03:19:29] <KitsuWhooa> It's from rst2man
[03:19:31] -!- bill-auger has joined #archlinux32
[04:41:57] -!- wjlafrance has quit [Quit: ZNC 1.8.2 - https://znc.in]
[04:42:16] -!- wjlafrance has joined #archlinux32
[04:55:57] -!- ConSiGno has quit [Remote host closed the connection]
[05:25:45] -!- abaumann has joined #archlinux32
[05:38:41] <abaumann> KitsuWhooa: this sounds like python-docutils
[05:38:48] <abaumann> haskell-generics-sop ebac3e044d7fe09f07ad8ecd48d6203c24f1e2c5 0000000000000000000000000000000000000000 extra curl: (22) The requested URL returned error: 500
[05:38:51] <abaumann> Download upstream-package "https://buildmaster.archlinux32.org/upstream-packages/haskell-generics-sop-ebac3e044d7fe09f07ad8ecd48d6203c24f1e2c5.tar.gz" from upstream via buildmaster cache failed. Curl exit code was 22.==> ERROR: PKGBUILD does not exist.
[05:38:55] <abaumann> "make_source_info" had empty output - eh, what?
[05:38:56] <abaumann> also the upstream cache has problems..
[05:41:12] <abaumann> but the file is there..
[05:42:18] <abaumann> Download upstream-package "https://buildmaster.archlinux32.org/upstream-packages/python-docutils-bb0bdf3442d032947fef6b2fb67332c528bbfaf1.tar.gz" from upstream via buildmaster cache failed. Curl exit code was 22.==> ERROR: PKGBUILD does not exist.
[05:42:22] <abaumann> "make_source_info" had empty output - eh, what?
[05:42:22] <abaumann> python-docutils bb0bdf3442d032947fef6b2fb67332c528bbfaf1 0000000000000000000000000000000000000000 1bb86318fd445e32004e9ffa7d1269018240434b community
[05:42:25] <abaumann> curl: (22) The requested URL returned error: 500
[05:42:29] <abaumann> welll.
[05:43:28] <abaumann> -rw-r--r-- 1 http http 1146 May 23 08:49 python-docutils-c62a1915d8f936a8bcba975876bf5f52ec4390f1.tar.gz
[05:43:31] <abaumann> -rw-r--r-- 1 http http 934 May 23 08:49 python-docutils-stubs-16af9d577f1e824ab3d172ea81a76edeed0b2a52.tar.gz
[05:44:10] <abaumann> now it works. mmh
[05:44:37] <abaumann> [master@buildmaster bin] master$ ~/force python-docutils >~/x; source ~/x python-docutils bb0bdf3442d032947fef6b2fb67332c528bbfaf1 0000000000000000000000000000000000000000 1bb86318fd445e32004e9ffa7d1269018240434b community
[05:44:41] <abaumann> curl: (22) The requested URL returned error: 500
[05:44:43] <abaumann> something.
[05:44:45] <abaumann> Download upstream-package "https://buildmaster.archlinux32.org/upstream-packages/python-docutils-bb0bdf3442d032947fef6b2fb67332c528bbfaf1.tar.gz" from upstream via buildmaster cache failed. Curl exit code was 22.==> ERROR: PKGBUILD does not exist.
[05:44:49] <abaumann> "make_source_info" had empty output - eh, what?
[05:44:52] <abaumann> aeh. ok.
[05:45:33] -!- abaumann has quit [Quit: leaving]
[06:15:09] -!- abaumann has joined #archlinux32
[06:17:37] <abaumann> So, all those packages from yesterdays sanity-check I moved by hand onto archive.archlinux32.org
[06:17:48] <abaumann> by hand == by bash script :-)
[06:18:15] <abaumann> I suspect a move went quite wrong
[06:18:22] <abaumann> in the buildmater
[06:47:35] <KitsuWhooa> :/
[06:47:45] <KitsuWhooa> Thanks for looking into it, abaumann
[06:55:31] <abaumann> -rw-r--r-- 1 http http 925706 Jul 9 07:49 python-docutils-1:0.20.1-1.0-any.pkg.tar.zst
[06:55:34] <abaumann> -rw-r--r-- 1 http http 310 Jul 9 07:49 python-docutils-1:0.20.1-1.0-any.pkg.tar.zst.sig
[06:57:15] <abaumann> despite all the errors and weirddness I see it seems I can actually force a build
[06:57:31] <abaumann> just get-package-updates is now back to 1 hour run time. :-(
[07:20:03] -!- abaumann has quit [Quit: leaving]
[09:40:32] <KitsuWhooa> Ugh...
[09:40:51] <KitsuWhooa> I wonder if there's a way to prevent it from running
[09:41:21] <KitsuWhooa> Until we are done building the massive queue of packages and sorting other build failures
[10:55:31] <KitsuWhooa> The Linux package is failing to build because ==> ERROR: Integrity checks (sha256) differ in size from the source array.
[13:23:56] -!- aimlesshk has joined #archlinux32
[13:24:49] <aimlesshk> I'm on a fast connection, downloaded an iso. Other than installing What else can I do while I'm on this fast connection? (it's a sneakernet question)
[13:25:39] <KitsuWhooa> Maybe download packages and their dependencies for offline installation. Things you will definitely need
[13:25:46] <KitsuWhooa> An iso is not enough to install an arch system offline
[13:25:57] <KitsuWhooa> If I understood this correctly anyway
[13:26:04] <KitsuWhooa> All packages are downloaded during installation
[13:27:12] <KitsuWhooa> If you have time you can install it using the fast connection in a VM and then write that VM image to an actual disk for use on an offline/slow Internet situation
[13:29:20] <aimlesshk> Ah good idea
[13:34:15] <aimlesshk> That'd be using the vboxmanage tool?
[13:52:53] <aimlesshk> Well, anyway, thanks! Packages are downloading so I'm definitely on the way with this. Many thanks :-)
[13:58:04] <aimlesshk> Package integrity failed. Any recommendations for other distros with 32bit support?
[14:17:21] -!- aimlesshk has parted #archlinux32
[15:00:16] <KitsuWhooa> <aimlesshk> Package integrity failed. Any recommendations for other distros with 32bit support? <-- I guess they gave up :p
[15:00:44] <KitsuWhooa> We really need to fix the constant mess of packages with keys not in the keyring or whatever
[15:00:56] <KitsuWhooa> Wouldn't be surprised if we need to spin a new iso with my key in it
[16:49:41] <girls> IMHO running get-package-updates less often solves nothing: It will simply run longer when you run it - the biggest part of the runtime is updating packages, and this needs time proportional to the amount of to-be-updated packages.
[16:53:07] <girls> lemme look into linux
[16:53:17] <girls> this probably just needs a bump
[16:53:31] <girls> i.e. it's a failsafe, so we adapt our config, when upstream changes theirs
[16:55:07] <girls> we could think of relieving the locking of get-package-updates a little, though
[16:57:38] <girls> ah, damn, my whole update-hash-diff-logic for linux* broke when changing to the state repository ... need to update devops/update-archlinux32-package
[17:02:11] <girls> Why do I only get 500's for https://buildmaster.archlinux32.org ?
[17:02:16] <girls> this sounds bad
[17:05:00] <KitsuWhooa> It will take a long time, yes, but at least some packages will get built that are otherwise broken because they were last built in 2020
[17:05:33] <girls> I don't see, how not-running get-package-updates helps with this or rather, how running it prevents it.
[17:09:34] <girls> re 500: my mistake - I tried to download the wrong commits (i.e. the ones from the state repository, not the ones *in* the state repository)
[17:24:00] <KitsuWhooa> If running it means the builders don't do anything for an entire hour...
[17:24:16] <KitsuWhooa> Along with other constant delays while trying to fetch jobs
[17:24:22] <KitsuWhooa> It adds up quickly imo
[17:24:22] <girls> if you don't run it, they will simply later not run for >1h
[17:24:37] <girls> I'll relieve the locks somewhat
[17:24:44] <girls> hopefully not-too-much breaks, then
[17:24:46] <girls> let's see :)
[17:25:08] <KitsuWhooa> Mmm :p
[17:28:24] <girls> the db actions are mostly atomar, anyways (it's not directly adding packages to the build-list, but to a to-be-decided list) - only deleting packages happens instantanous - but hey, they're deleted anyways :D
[17:29:44] <girls> the worst, I can think of, currently is, that the dependency tracking of the new packages is broken, as long as the script is not done yet - in combination with removing the superseeded package instantanously, this may lead to funny effects.
[17:30:50] <girls> i.e. a depends on b, both are on the build-list, b gets an update, while a slave requests work. Then it might be, that the old b is already removed, but the new b is not yet properly tracked - then a might get handed out prematurely (or it might be ignored completely, which would be good - I'm not sure).
[17:31:27] <KitsuWhooa> Hmmm
[17:31:37] <girls> but hey, still better than blocked slaves >:-)
[17:33:06] <KitsuWhooa> I still get messages every now and then that my slaves are building outdated packages
[17:35:57] <girls> that's not really solvable
[17:36:09] <girls> just restart the slave in these cases
[17:36:35] <girls> unfortunately, the arch devtools build via sudo, so one would need to kill them as root, too - so we cannot automate this
[17:39:04] <girls> one could allow to have two versions on the build-list, iff the older one is currently being built - but that would be a quite complex change, because we would have to drop a unique constraint and what we expect from the table
[17:40:29] <girls> also build slaves do not necessarily tell, when they stop building a package (think of power failure or alike)
[17:52:46] <KitsuWhooa> Ah I see
[17:52:52] <KitsuWhooa> Fair enough
[17:53:28] <girls> if it becomes a big pain point, we can think of implementing something, but currently I would see it "only" as nice-to-have
[17:59:36] <KitsuWhooa> Yeah, of course
[18:55:17] -!- finsternis has joined #archlinux32
[19:30:42] <girls> ok, finally I fixed the kernel update script - new linux and linux-zen should be on their way
[19:31:10] <girls> @abaumann: other kernels will need fixes in the PKGBUILD, once upstream bumps them - they replaces sha256sum by b2sum ...
[20:07:38] <KitsuWhooa> awesome
[20:07:42] <KitsuWhooa> thanks!
[23:11:41] -!- T`aZ has quit [Ping timeout: 245 seconds]
[23:24:43] -!- T`aZ has joined #archlinux32