[03:55:41] <legoktm>	 !log codesearch restarting everything to add new "wmcs" instance
[03:55:44] <stashbot>	 Logged the message at https://wikitech.wikimedia.org/wiki/Nova_Resource:Codesearch/SAL
[10:48:55] <Wurgl>	 Is there some image with mono in toolforge-jobs images ?
[13:56:27] <taavi>	 Wurgl: not yet. can you create a phabricator task about that?
[13:58:30] <Wurgl>	 okay. Just wondering if I am the only one
[14:06:25] <Wurgl>	 https://phabricator.wikimedia.org/T305780 <-- is that enough?
[14:32:19] <Krinkle>	 bd808: I noticed the redirectfixer on officewiki marks edits as minor, which is great (in light of T305734) - I'm looking at pywiki source and not seeing where fix_redirects does this. Does it use something else?
[14:32:19] <stashbot>	 T305734: Page move notification e-mails sent for watchlisted pages even when actor has the flood right - https://phabricator.wikimedia.org/T305734
[14:32:38] <Krinkle>	 In any event, I figured we can make core do the same. it already skips RC but not yet minor. https://gerrit.wikimedia.org/r/c/mediawiki/core/+/778617
[14:33:45] <AntiComposite>	 I believe the default in pywikibot, for any edit, is bot and minor unless otherwise specified
[14:42:33] <Krinkle>	 AntiComposite: ah that would do it, okay :)
[15:29:05] <taavi>	 Wurgl: why do you need both im the same container image? having multiple runtimes in the same container is planned but not currently supported
[15:30:12] <taavi>	 !log tools manually prune user.log on tools-prometheus-03 to free up some space on /
[15:30:14] <stashbot>	 Logged the message at https://wikitech.wikimedia.org/wiki/Nova_Resource:Tools/SAL
[15:32:30] <Wurgl>	 taavi: There is a c#-binary which does the computing and the output is processed with php
[15:32:56] <Wurgl>	 The c#-binary can also be used standalone with a gui
[15:33:29] <Wurgl>	 Historical reasons. The guy who wrote the program/started the tool is inactive since 2017
[16:31:24] <AmandaNP>	 So can I use cron for the new Kubernetes? Also can I specify a limit of how long a job can run?
[17:24:57] <JJMC89>	 AmandaNP: you can use cron via --schedule. there isn't a way to limit the run duration.
[17:27:06] <wm-bb>	 <lucaswerkmeister> I imagine `--command 'timeout 10m ...'` would work to kill the command if it takes too long
[17:31:41] <JJMC89>	 I think that will cause the job to be retried -- since it will cause a failure
[17:32:33] <AntiComposite>	 you could certainly do it with activeDeadlineSeconds if you felt like writing some yaml
[17:32:47] <wm-bb>	 <lucaswerkmeister> JJMC89: good point
[17:40:20] <AntiComposite>	 looks like jobs and cronjobs have a RestartPolicy: Never, so that shouldn't be an issue
[17:46:21] <JJMC89>	 they have backoffLimit: 1, so they get retried once
[17:50:18] <AntiComposite>	 timeout --preserve-status ?
[17:52:21] <wm-bb>	 <lucaswerkmeister> well, the command will presumably exit nonzero when it gets killed, so preserving that wouldn’t help as much
[17:52:35] <AntiComposite>	 yeah, you'd have to intercept that
[17:52:37] <wm-bb>	 <lucaswerkmeister> but if the command is run by a shell (I think it is?) then you could just `; exit 0` at the end I guess
[17:53:28] <AntiComposite>	 I'd suggest filing a phab task saying you want to set a timeout
[17:54:01] <taavi>	 is backoffLimit: 1 a k8s default or something toolforge-jobs sets explicitely?
[17:54:26] <JJMC89>	 we have it set explicitly - not sure what the k8s default is
[17:54:30] <AntiComposite>	 https://phabricator.wikimedia.org/diffusion/CTJF/browse/main/tjf/job.py$268
[17:54:39] <AntiComposite>	 default is 6
[18:57:09] <AmandaNP>	 and I've officially messed up my venv...i'm so confused and lost rn. I've broken my bot that runs on python 3
[19:05:57] <AmandaNP>	 I think I ended up fixing it
[19:25:09] <andrewbogott>	 !log admin reimaging cloudbackup1002-dev to bullseye
[19:25:12] <stashbot>	 Logged the message at https://wikitech.wikimedia.org/wiki/Nova_Resource:Admin/SAL
[19:37:42] <taavi>	 !log admin add 'puppet-enc' service & endpoint to keystone T274666
[19:37:46] <stashbot>	 Logged the message at https://wikitech.wikimedia.org/wiki/Nova_Resource:Admin/SAL
[19:37:47] <stashbot>	 T274666: Add keystone auth middleware to the puppet enc api - https://phabricator.wikimedia.org/T274666
[19:55:21] <andrewbogott>	 !log admin reimaging cloudbackup1001-dev to bullseye
[19:55:24] <stashbot>	 Logged the message at https://wikitech.wikimedia.org/wiki/Nova_Resource:Admin/SAL
[22:08:39] <dschwen>	 Is everybody having fun migrating their tools to kubernetes? (I'm not)
[22:10:06] <dschwen>	 Bootstrapping a python venv for kubernetes is a huge pain. I need py3exiv2 in one tool, and that does not build without prerequisites. And to build the required exiv2 library I need cmake, which doesn't exist on the python 3.9 image... :-(
[22:13:20] <dschwen>	 another tool is using lighttpd with cgi. None of the kubernetes worked. I ended up migrating to gridengine buster
[22:13:48] <dschwen>	 I have half a dozen more tools and I now expect each one to give me some surprises
[22:14:52] <dschwen>	 I found it not very easy to even find out what the available images are (not to mention what software they offer)
[22:15:46] <dschwen>	 There are some example for `jsub` replacements, but no mention of how to replace the `--once` option (which I found extremely useful). Do I have to go back to implement locking myself now?