| 8 Feb 2022 |
 cole-h | teleports behind Hydra's alt inputs and GitHubStatus plugin "Nothing personnel, kid." | 04:57:36 |
 @grahamc:nixos.org | das_j: I was looking at https://keepachangelog.com/en/1.0.0/ and they really want you to be having _released software_. I wonder what format would work well for our use case. | 15:16:57 |
| @grahamc:nixos.org | Some complications are merge order and authorship date are not relevant when the merge moment is. | 15:18:25 |
 das_j | hm not sure if releases make a lot of sense for the current state of hydra | 15:18:59 |
| @grahamc:nixos.org | same | 15:19:10 |
| @grahamc:nixos.org | Daily releases =) | 15:19:50 |
| das_j | Why keep yourself back? Github action to tag a new release on every merge :P | 15:30:00 |
| @grahamc:nixos.org | I mean | 15:30:36 |
| @grahamc:nixos.org | why not? | 15:30:39 |
| das_j | why even bother with releases then? | 15:33:55 |
| @grahamc:nixos.org | =) | 15:37:32 |
 ajs124 | follow the minio model. release every Friday (I think). | 15:53:33 |
 hexa | let fools deploy over the weekend | 16:05:03 |
| hexa | * let fools crash and burn over the weekend | 16:05:12 |
| @grahamc:nixos.org | there you go | 17:06:01 |
| @grahamc:nixos.org | https://discourse.nixos.org/t/dropping-jobsetinputalts/17584 | 17:55:16 |
| 9 Feb 2022 |
 Amanda (she/her) | Anybody else have hydra seem to slowly be leaking postgres handles? It seems to slowly build up, and I'm not even sure which service it is that it's happening from | 00:14:41 |
| Amanda (she/her) | Eventually stuff starts falling over because postgres is like "No. Stop." | 00:15:02 |
| das_j | Maybe we are not running into it because we are not running postgres with default settings… | 00:16:33 |
| Amanda (she/her) | To be specific, this is what I start to see, though it's not always hydra-gc-update-roots:
MESSAGE DBIx::Class::Storage::DBI::catch {...} (): DBI Connection failed: DBI connect('dbname=hydra;user=hydra;','',...) failed: FATAL: remaining connection slots are reserved for non-replication superuser connections at /nix/store/12d5ignyfdq5kpdj2hk5yympd9rdkw49-hydra-perl-deps/lib/perl5/site_perl/5.34.0/DBIx/Class/Storage/DBI.pm line 1639. at /nix/store/5bqzhwky888dyfyv133ak4nzqfaa98bx-hydra-2021-08-11/bin/.hydra-update-gc-roots-wrapped line 72
| 00:17:23 |
| das_j | ah yeah looks like we have max_connections = 250; set | 00:17:42 |
| Amanda (she/her) | I think the last time this crept up on me it was hydra-queue-runner which started falling over due to it | 00:18:34 |
| @grahamc:nixos.org | Amanda (she/her): can you list open connections? they should be identified which "app" they're coming from | 00:33:00 |
| ajs124 | In reply to @janne.hess:helsinki-systems.de
Maybe we are not running into it because we are not running postgres with default settings… we used to, before we adjusted the settings. remember the nice error message catalyst always throws up? the one that's translated to austrian, for some reason? | 00:36:58 |
| @grahamc:nixos.org | lmao | 00:37:12 |
| ajs124 | 
Download image.png | 00:37:49 |
| ajs124 | that's our hydra host. don't think that db is used for anything else. | 00:38:08 |
| @grahamc:nixos.org | it may be the webserver holding open connections one in each worker | 00:38:27 |
| @grahamc:nixos.org | it'd be interesting for you to list the app of each open connection :) | 00:38:49 |
| @grahamc:nixos.org | ("app" is a feature of postgres) | 00:38:58 |
