| 19 Nov 2023 |
|  pbsds changed their display name from pbsds to pbsds (federation borken, may not see reply). | 03:35:20 |
 vcunat | In reply to @marius851000:mariusdavid.fr
Hello. I set up an Hydra server for doing some custom CI of a package manager (portmod), and I had a wrongly written fetch script for a fixed output derivation. I now fixed it, but Hydra doesn’t re-evaluate it (certainly cause the result set end up having the same hash). Is there some way I can force it to rebuild all failed derivation according to a new definition even if the result set hash is identical?
(I already had this problem once, and it seems changing the hash of the dependant and re-evaluating wasn’t enought, but I had to change the name anyway, which resulted in a different hash and so a rebuild for every sources) I'm late but yes, that's an issue. I know nothing better than changing the derivation somehow - in a way that changes the output path, so e.g. the name of the fixed-output derivation. | 08:15:29 |
| vcunat | Or you could like create a different jobset that contains the corrected derivation - that way you get a new job even if the output path is exactly the same. | 08:16:43 |
 Rick (Mindavi) | In reply to @raitobezarius:matrix.org
I don't think you can delete jobsets in Hydra Open a javascript console while logged in and run deleteJobset or sth like that while on the jobset page | 10:55:20 |
| Rick (Mindavi) | The UI doesn't expose it, presumably because it's a very expensive operation on bigger instances | 10:56:08 |
|  ZXGU joined the room. | 10:56:50 |
 magic_rb | DBIx::Class::Storage::DBI::_execute(): Wide character in subroutine entry at /nix/store/z2f0zs45kx5r2x9i1yw7j7faawhbfvbb-hydra-perl-deps/lib/perl5/site_perl/5.36.0/DBIx/Class/Storage/DBI.pm line 2013. at /nix/store/6iszykmg0zqf83gj8mdy197pkf645lyj-hydra-0.1.20230803.a5ff881/bin/.hydra-eval-jobset-wrapped line 729
{UNKNOWN}: Wide character in subroutine entry at /nix/store/z2f0zs45kx5r2x9i1yw7j7faawhbfvbb-hydra-perl-deps/lib/perl5/site_perl/5.36.0/DBIx/Class/Storage/DBI.pm line 2013. at /nix/store/z2f0zs45kx5r2x9i1yw7j7faawhbfvbb-hydra-perl-deps/lib/perl5/site_perl/5.36.0/Catalyst/Model/DBIC/Schema.pm line 526
evaluation of jobset ‘dotfiles:flake-update (jobset#16)’ failed with exit code 255
I've had this problem for ages where evaluation output wouldnt be visible in the webui, only now am i getting to actually fixing it. Does that look familiar to anyone? there are a few issues open and closed related to UTF-8 encoding but nothing specifically with the database
| 12:18:51 |
| pbsds changed their display name from pbsds (federation borken, may not see reply) to pbsds. | 20:38:39 |
|  PaulZ (pfz4) joined the room. | 20:46:47 |
| PaulZ (pfz4) changed their display name from Paul / pfz4 to PaulZ (pfz4). | 20:48:37 |
| 20 Nov 2023 |
|  Paul-Nicolas Madelaine joined the room. | 10:33:19 |
| 21 Nov 2023 |
|  niksnut changed their display name from niksnut to Eelco. | 16:37:37 |
 @antifuchs:asf.computer | is there a way to configure hydra to load a user-defined plugin, or do I have to build/patch a new hydra package for that? | 22:14:15 |
| @antifuchs:asf.computer | (sounds like you can set PERL5LIB to set the perl module search path; if the module is in the right namespace, it might work? I'll try it out (-:) | 22:20:55 |
| @antifuchs:asf.computer | got a module that gets loaded & called when various eval hooks are called. This is gonna be a very fun time! | 22:58:17 |
| 22 Nov 2023 |
| @antifuchs:asf.computer | oh man, so the evalStarted/evalFailed/etc hooks get a tmpId for the evaluation that is not available in the database? what good is that )-: | 14:22:48 |
| @antifuchs:asf.computer | I think I can at least dissect that temporary ID on evalFailed to find a jobsetevals entry corresponding to the jobset that failed, which would let me set a github status on that eval's corresponding commit. but, damn, this would be much easier if there was a hydra-side ID that gets threaded through everywhere /: | 14:30:29 |
| @antifuchs:asf.computer | * oh man, so the evalStarted/evalFailed/etc hooks get a tmpId for the evaluation that is not available in the database (and database IDs that aren't available on any hook)? what good is that )-: | 14:30:47 |
| @antifuchs:asf.computer | oh wait, evalAdded and evalCached do get the evaluation ID. too bad about evalFailed though... | 14:39:19 |
| 23 Nov 2023 |
| @antifuchs:asf.computer | 
Download image.png | 14:54:52 |
| @antifuchs:asf.computer | oh gossssshhhhhhh it works, I finally have a plugin that posts a github commit status if evaluation of a flake commit fails | 14:54:52 |
| @antifuchs:asf.computer | would of course love a thing that lets me drop a "evaluation of this commit started" status, instead of just reporting the end of the evaluation but that is not how hydra hydra's | 14:56:06 |
 raitobezarius | crazy | 16:01:06 |
| @antifuchs:asf.computer | I’m so happy, this puts an end to me waiting for builds to finish that would never get started (: | 20:06:46 |
| 24 Nov 2023 |
| @antifuchs:asf.computer | here's the code I use to get that commit status on my nixos config repo: https://gist.github.com/antifuchs/5c2394eb88e0ca436ba9113e2aaeaed7 | 15:43:48 |
| 25 Nov 2023 |
|  edgar.vincent joined the room. | 00:26:36 |
| magic_rb | im finally not dying of covid so im debugging, since i bumped my flake recently hydra-server (starman, the www part) just doesn't start
/ # HOME=~hydra-www chpst -b hydra-server -u hydra-www:hydra /nix/store/grgqdxx6z3jpbxdmfkaam9cdg485jj56-hydra-env/bin/hydra-server hydra-server -f -h '*' -p 3000 --max_spare_servers 5 --max_servers 25 --max_requests 100 -d
[debug] Debug messages enabled
[debug] Statistics enabled
[debug] Loaded Config "/var/lib/hydra/hydra.conf"
[debug] Setting up auth realm dbic
[debug] Session Store file: /var/lib/hydra/www/session_data
[info] Plugin::PrometheusTiny metrics endpoint installed at /metrics
any idea how i could debug this further? it just does nothing rn
| 16:06:01 |
| magic_rb | looking at my commits, it was indeed the flake lock bump | 16:22:44 |
| magic_rb | ah but im not following upstream | 16:23:25 |
| magic_rb | scratch that, not using a random fork of hydra helps | 16:57:41 |
