#archlinux32 | Logs for 2023-03-10
Back
[01:34:54] -!- drathir_tor has quit [Remote host closed the connection]
[01:35:24] -!- drathir_tor has joined #archlinux32
[06:55:50] -!- abaumann has joined #archlinux32
[06:55:50] <buildmaster> Hi abaumann!
[06:55:50] <buildmaster> !rq abaumann
[06:55:51] <phrik> buildmaster: <abaumann> "PAM starts to hit the fan." :-)
[06:56:44] <abaumann> aha. I see a get-package-updates since March 9th running on the buildmaster and awk tells me in ps -a "input contains a loop" which doesn't sound like something which would eventually finish maybe? :-)
[07:10:03] <abaumann> ghc: "Currently only 9 is supported. System LLVM version: 14.0.6
[07:10:28] <abaumann> As far as I know LLVM you actually push your luck here whether the compiler even works.
[07:10:36] <abaumann> But upstream is doing it that way.. so..
[07:10:49] <abaumann> ..it might just work.. :-)
[07:56:53] <abaumann> ok, I'm going to kill now this endless loop in get-package-updates
[08:36:33] -!- eschwartz has quit [Ping timeout: 255 seconds]
[08:49:04] -!- eschwartz has joined #archlinux32
[16:06:41] <girls> get-package-update shouldn't really run that long - do you have any clue, what it was doing?
[16:07:40] <girls> just great: the journal is truncated right before you killed it - I cannot see, what it did last :(
[16:08:17] <girls> but if the buildmaster is not insane, the worst could be, that we missed some updates from upstream (but should not happen, actually)
[16:11:50] <abaumann> oups. Sorry for killing it. Maybe there was a cycle somewhere..
[16:12:01] <abaumann> ..since then it works without issues..
[16:12:27] <girls> no, I think, killing was the right thing - it's just sad, that I cannot see anymore, what it was trying to do
[16:12:46] <girls> do you have any hint, what it was doing?
[16:12:54] <abaumann> maybe I moved tmp.* files into the work/broken subdir?
[16:13:06] <abaumann> tsorting and awking some packages, I presume..
[16:13:49] <girls> there's no more tsort anymore
[16:13:55] <girls> it's all in mariadb
[16:14:00] <girls> (that's, why it's so slow :D)
[16:14:00] <abaumann> mmh. interesting..
[16:14:14] <girls> maybe some PKGBUILD is broken and hangs indefinitely when sourced
[16:14:16] <girls> hmmm
[16:14:28] <abaumann> there is one tsort in ../lib/mysql-functions..
[16:14:37] <girls> it is? hmm
[16:14:48] <abaumann> input contains a loop
[16:14:57] <abaumann> there is also the awk script just below.
[16:14:58] <girls> ah, to find the loops, right
[16:15:05] <abaumann> mysql_find_build_assignment_loops
[16:15:14] <girls> yeah, I see it
[16:15:20] <abaumann> cool. so we know where.
[16:15:28] <abaumann> maybe it was just taking a _long_ time..
[16:15:39] <abaumann> ..and I was inpatient.. as usual. :)
[16:15:50] <girls> yeah, the script is without branches, it cannot have an infinite loop
[16:16:02] <girls> well, >1d runtime seems long in any case ...
[16:17:19] <girls> ok, so it's just in the "Aftermath" step, that you killed it - those are uncritical, because they should be idempotent and also recover from any state
[16:17:47] <girls> i.e. one coule run mysql_find_build_assignment_loops at almost any time manually
[16:19:12] <abaumann> uff. good news. :-)
[16:19:43] <girls> well, we can recover from almost any state ;)
[16:23:27] <abaumann> mmh. interesting. usually I see one mysqld 100% cpuish, now there are 3 x 100%
[16:23:38] <girls> no idea
[16:23:52] <abaumann> so, some paralellism kicking in, maybe..
[17:39:39] -!- abaumann has quit [Quit: leaving]
[20:08:23] -!- drathir_tor has quit [Remote host closed the connection]
[20:09:49] -!- drathir_tor has joined #archlinux32