[06:00:34] !log tools.mjolnir upgraded to https://github.com/matrix-org/mjolnir/releases/tag/v1.1.20, switched to node12 image [06:00:35] Logged the message at https://wikitech.wikimedia.org/wiki/Nova_Resource:Tools.mjolnir/SAL [13:02:22] !log tools.quickcategories deployed 0213175db9 (migrate to dataclasses and abc) [13:02:25] Logged the message at https://wikitech.wikimedia.org/wiki/Nova_Resource:Tools.quickcategories/SAL [14:04:55] I’m trying to hack on the webservice script on Toolforge, but having trouble getting a copy to run [14:05:04] does anyone have experience with this? [14:05:53] I cloned tools-webservice.git into the tool home, but when I run scripts/webservice, it still uses the globally-installed toolsws/ instead of the one in the git clone [14:06:33] symlinking toolsws/ into scripts/ fixes that, but then it apparently can’t talk to k8s anymore, or at least the status output changes to “your webservice is not running” [14:07:54] `python3 setup.py develop --prefix ~/.local` doesn’t help either, seems to run the globally-installed toolsws/ again [14:13:12] huh, /usr/bin/webservice is also pretty outdated? from August 15, while there have been more commits since then [14:14:15] hah, but once I check out 34d8855358, scripts/webservice works \o/ [14:14:32] so it’s not python setup.py weirdness, it’s just broken on master 🤷 [14:27:52] lucaswerkmeister: https://wikitech.wikimedia.org/wiki/Portal:Toolforge/Admin#Testing/QA_for_a_new_toollabs-webservice_package [14:28:43] thanks – that looks like it would replace the symlinking of toolsws? [14:29:15] it tells python to use the local toolsws/ directory instead of the global one [14:29:19] ok, got it [14:29:44] also, https://gerrit.wikimedia.org/r/c/operations/software/tools-webservice/+/637813 breaks compat between versions before it unless a migration script is run (which we will do when deploying the next update) [14:30:23] we deploy it as a debian package, so the latest commit to debian/changelog usually tells you what is live and what's not [14:31:35] out of curiosity, what are you hacking in it? [14:35:30] https://phabricator.wikimedia.org/T290833 :) [14:35:40] and I think I just got a working version so I’ll probably upload it to Gerrit shortly [14:52:31] @seen Pathoschild [14:52:57] hmm did the command format change? [15:06:37] zhuyifei1999_: https://meta.wikimedia.org/wiki/Wm-bot#Private_message_commands ? [15:06:48] (tl;dr - send a PM?) [15:06:58] (yeah I figured) [15:07:09] it used to work publicly tho [15:07:16] thanks anyways [15:07:33] yeah I recall it working publicly too *shrug* [15:34:56] it's configurable per-channel [15:35:36] majavah: webservice hack is on Gerrit if you want to take a look https://gerrit.wikimedia.org/r/c/operations/software/tools-webservice/+/721989 [15:37:02] ooh, thanks! [15:40:31] lucaswerkmeister: left a quick comment, I'm afraid that I'm rather busy in the next few weeks and might not be able to do an in-depth review [15:41:00] eek, I didn’t know about those replicasets [15:41:09] * lucaswerkmeister frowns at kubectl [15:41:24] otherwise: that’s okay, it’s not like I need this urgently – I just had some time today and thought I’d try to implement it [15:43:23] I guess it’s sorta intentional, because you’re meant to be able to roll back to those replicasets if the new one turns out to be bad? [15:43:35] (in general – here of course the rollback would be pointless if the only difference is a timestamp annotation) [15:45:15] but they should be cleaned up at some point [15:56:27] oh wow, kubectl needs tons of threads apparently? [15:56:55] I’m just trying to run two `kubectl get`s in parallel `watch`es, and they alternatingly run out of processes/threads :( [15:57:42] unfortunately yeah :/ [15:58:32] anyways I deleted the stale replicasets now (manually) [15:58:37] since I created so many with those tests