#archlinux32 | Logs for 2023-06-09

Back
[00:18:38] -!- drathir_tor has quit [Remote host closed the connection]
[00:24:07] -!- drathir_tor has joined #archlinux32
[05:03:37] -!- T`aZ has quit [Ping timeout: 240 seconds]
[05:06:12] -!- T`aZ has joined #archlinux32
[05:27:50] -!- abaumann has joined #archlinux32
[05:27:50] <buildmaster> Hi abaumann!
[05:27:50] <buildmaster> !rq abaumann
[05:27:51] <phrik> buildmaster: <abaumann> haskell-base-compat-batteries.. ah.. Haskell comes with batteries included.. in constrast to my Arduino box..
[05:27:55] <abaumann> gcc:
[05:27:57] <abaumann> gzip: stdin: unexpected end of file
[05:27:58] <abaumann> tar: Child returned status 1
[05:27:58] <abaumann> tar: Error is not recoverable: exiting now
[05:28:00] <abaumann> ==> ERROR: pkgrel is not allowed to be empty.
[05:28:02] <abaumann> ==> ERROR: pkgver is not allowed to be empty.
[05:28:05] <abaumann> ==> ERROR: pkgname is not allowed to be empty. failed. failed. failed. failed.
[05:28:08] <abaumann> Done.
[05:28:29] <abaumann> this is because the merging of PKGBUILDs is failing. What I see is that gcc's PKGBUILD consists of the diff PKGBUILD only without the upstream PKGBUILD
[06:27:44] -!- drathir_tor has quit [Ping timeout: 240 seconds]
[06:35:06] -!- drathir_tor has joined #archlinux32
[07:08:21] -!- abaumann has quit [Remote host closed the connection]
[07:28:55] -!- abaumann has joined #archlinux32
[07:28:55] <buildmaster> Hi abaumann!
[07:28:55] <buildmaster> !rq abaumann
[07:28:56] <phrik> buildmaster: <abaumann> If it takes long.. it works..
[07:37:16] <abaumann> aha, I see a failing gzip after a curl fetch upstream..
[07:37:49] <abaumann> -rw-r--r-- 1 http http 0 Jun 9 07:23 gcc-9c162ff08cf4d0c1bb98a7d72b0b46ec949890ed.tar.gz
[07:37:56] <abaumann> yep, that might be hard to build :-)
[07:38:06] <abaumann> I have to fix our upstream gitlab fetcher..
[07:40:18] <abaumann> aha, deleting the file and refetching still results in a 0 byte file
[07:41:03] <abaumann> 127.0.0.1:9050 huh
[07:41:08] <abaumann> that one run away..
[07:41:39] <abaumann> our tor gateway was not enabled. :-)
[07:42:32] <abaumann> -rw-r--r-- 1 http http 17946 Jun 9 09:42 gcc-9c162ff08cf4d0c1bb98a7d72b0b46ec949890ed.tar.gz
[07:42:35] <abaumann> much better
[07:42:53] <abaumann> I'll add a sanity check to the fetch package php script..
[07:43:21] <abaumann> long term we should really find a better solution to access upstreams gitlab than going through tor.
[07:48:10] <abaumann> the builmaster on /srv/http has a lot of unchecked in php scripts, cpies and un-gitted stuff, I'll deal with it - the current mess is quite likely my doing :-)
[07:48:32] * abaumann thinks his trusty 15 year old laptop has a keyboard issue..
[07:48:50] <abaumann> (random failing missing characters)
[08:07:47] <abaumann> mmh. packages are still only building with some straws..
[08:13:39] <abaumann> https://buildmaster-status.archlinux32.org
[08:13:50] <abaumann> the failure rate is way too high, far too many packages are not building..
[08:15:58] <T`aZ> (btw, I'm getting a double 403 error on that page)
[08:21:48] <abaumann> oups. https://buildmaster.archlinux32.org
[08:21:50] <phrik> Title: Buildmaster for Archlinux32 packages (i486, i686, pentium4, any) (at buildmaster.archlinux32.org)
[08:22:08] <abaumann> that's the correct one. The other one with -status is kinda - aeh - private. :-)
[08:55:42] <abaumann> I think, I'll open the git:// protocol read-only on the buildmaster, pkgutil needs a shallow checkout of archlinux32 diffs, and the https way is just a little bit too restrictive
[08:55:58] <abaumann> on the slaves we can go over ssh:// when we do something fancy
[09:21:10] <abaumann> surely some security companies will continue to spam me and tell me that git repositories are accessible to the public..
[09:21:18] <abaumann> ..well, that's the idea with public git repos :-)
[09:49:05] <abaumann> mmh. I guess modification git revision 0000000000000000000000000000000000000000 means we don't have local patching and build from PGBUILD upstream directly..
[09:49:37] -!- abaumann has quit [Quit: leaving]
[10:06:13] -!- T`aZ has quit [Ping timeout: 250 seconds]
[10:08:21] -!- T`aZ has joined #archlinux32
[10:38:14] -!- swimm3r has quit [Ping timeout: 246 seconds]
[11:24:44] -!- swimm3r has joined #archlinux32
[14:11:44] -!- eschwartz has quit [Ping timeout: 248 seconds]
[15:22:33] -!- T`aZ has quit [Ping timeout: 250 seconds]
[15:24:27] -!- T`aZ has joined #archlinux32
[16:14:41] -!- abaumann has joined #archlinux32
[16:14:41] <buildmaster> Hi abaumann!
[16:14:42] <buildmaster> !rq abaumann
[16:14:42] <phrik> buildmaster: <abaumann> this is not a database.. this is pure stupidity..
[16:14:42] <abaumann> error: RPC failed; HTTP 408 curl 22 The requested URL returned error: 408
[16:15:15] <abaumann> now with glibc. This starts to really bother me. Ommitting the --mirror didn't help here, setting http.postBuffer neither..
[16:15:19] -!- abaumann has quit [Client Quit]
[16:26:56] -!- drathir_tor has quit [Ping timeout: 240 seconds]
[16:53:16] -!- drathir_tor has joined #archlinux32
[18:31:12] <girls> KitsuWhooa: your key is now in the keyring, you should be able to build and upload packages, now :)
[18:31:24] <girls> btw: thank you for sharing the cpu cycles!
[18:35:32] <girls> Foxboron: the state repo is broken: "extra-x86_64/arbtt" has no git hash, but a pkgver three times
[18:36:56] <girls> umm, and some haskell-* packages show the same issue
[18:39:49] <KitsuWhooa> girls: 21:39:34 <buildmaster> Your buildslave "build-tasossah" uploaded a package with a signature of a key unknown to the mysql database:
[18:39:51] <KitsuWhooa> 21:39:34 <buildmaster> NULL glibc-2.37-3.3-i686.pkg.tar.zst
[18:40:25] <girls> ah, the mysql database
[18:40:27] <girls> gimme a sec
[18:46:23] <girls> KitsuWhooa: can you try again?
[18:46:43] <KitsuWhooa> once glibc builds again
[18:46:49] <girls> :-/
[18:47:14] <girls> I thought, build-packages would get a temporary error and just retry the upload, hmmmm
[18:47:24] <KitsuWhooa> it went straight back to cloning and trying again
[18:47:26] <KitsuWhooa> so I stopped it
[18:47:44] <girls> ok, maybe I didn't consider a key error to be solavble on the buildmaster side
[18:51:00] <girls> abaumann, FYI: I "fixed" the state-repo issue by explicitly ignoring packages of the wrong form (i.e. "$pkgbase $pkgver $pkgver $pkgver" instead of "$pkgbase $pkgver $pkgver $commit"), because we cannot build them anyways - and else, we're blocked with updates completely. However, this has the side effect of marking those packages as to-be-deleted. But TBH: I don't care.
[18:51:46] <Foxboron> girls: hm, interesting
[18:55:03] <Foxboron> girls: thanks for the note, we'll try figure it out
[18:55:06] <KitsuWhooa> girls: https://tasossah.com
[18:55:07] <phrik> Title: build stuck uploading (at tasossah.com)
[18:55:21] <girls> Foxboron: thanks for looking into it :)
[18:55:29] <Foxboron> girls: I assume it's a bash quirk somewhere
[18:55:51] <girls> KitsuWhooa: yes, that's normal, the db is currently locked due to package updates
[18:55:56] <KitsuWhooa> ah
[18:57:51] <girls> the upstream state repo was/is broken and I just fixed our parse script to ignore the errors - thus, there is now a longish update running ...
[18:58:04] <girls> the goal was to get the recent archlinux32-keyring into the arch32 repositories
[18:58:11] <girls> (currently, it's only in [releng])
[18:59:51] <girls> abaumann: another proposal: I would remove archlinux32-keyring-transition. It's annoying to get it updated regularly, it's currently quite outdated and thus not really helpful. Anyone, wanting to "transition" from arch to arch32 will have to do some effort and build up some initial trust anyways.
[19:00:50] <girls> Foxboron: but the good news (for us) is, that it's not intended - otherwise I would have to rewrite quite some parts of the buildmaster (again) :D
[19:02:54] <bill-auger> anyone transitioning from arch to arch32 today, has not upgraded their system in a dog's age - i would recommend re-installing fresh anyways
[19:03:11] <girls> yeah, exactly
[19:24:41] -!- eschwartz has joined #archlinux32
[19:49:15] <girls> KitsuWhooa: looks, like it's working, now :)
[20:29:07] <KitsuWhooa> girls: still seems stuck at the same place
[20:34:51] <girls> but you already successfully built and uploaded glibc for i686 :)
[20:34:58] <girls> ... update is still running ...
[20:35:17] <KitsuWhooa> oh it did get uploaded?
[20:35:21] <KitsuWhooa> cool!
[20:35:24] <KitsuWhooa> I didn't realise
[20:37:38] <girls> yeah, it's not easily noticable from "the outside", but the database says, it's yours
[20:39:33] <girls> ah, damn - you're right
[20:40:07] <KitsuWhooa> :p
[20:40:22] <girls> hmm, or it's too late for me to do sql ...
[20:40:28] <girls> KitsuWhooa | glibc-2.37-3.3-i686.pkg.tar.zst
[20:40:36] <girls> I thought, the package on the mirror said the same
[20:42:00] <girls> for whatever reason, the buildmaster saved it as "compression = NULL", so the default - and not explicitly as "compression = zst" .... strange
[20:44:29] <girls> congratulation: you triggered a (minor) bug :)
[20:45:53] <KitsuWhooa> :D
[20:45:56] <KitsuWhooa> I'm cursed
[20:51:00] <girls> ok, fixed it manually - maybe this is fallout from the transition, and you were just lucky enough, that your first built package did have this glitch in the database :D
[20:51:08] <girls> (abaumann had 12 such packages - I had none)
[20:51:13] <KitsuWhooa> we'll see
[20:51:50] <girls> now for the hunt to the missing link in https://buildmaster.archlinux32.org
[20:52:16] <KitsuWhooa> hah
[21:02:56] <girls> there you go :)
[21:52:59] -!- drathir_tor has quit [Remote host closed the connection]
[22:23:52] -!- drathir_tor has joined #archlinux32
[22:29:58] <KitsuWhooa> finally we're building
[22:29:59] <KitsuWhooa> wooo
[22:31:20] -!- drathir_tor has quit [Ping timeout: 240 seconds]
[22:34:12] -!- drathir_tor has joined #archlinux32
[22:59:49] <KitsuWhooa> Build master is not sane.
[22:59:51] <KitsuWhooa> didn't take long
[23:17:36] -!- drathir_tor has quit [Remote host closed the connection]
[23:28:04] -!- drathir_tor has joined #archlinux32