[08:23:15] elukey: got your email 👍 [08:24:58] arturo: o/ sorry for the spam, didn't see you online! [08:25:26] no problem! [08:25:52] out of curiosity, does irccloud mark me offline somehow when I'm not staring at the chat browser tab? [08:30:55] irccloud has an Auto-away feature enabled by default [08:30:59] "Auto-away Set your away status when you leave the page" [08:42:31] I see, but it's unchecked in my case, i.e disabled [08:42:44] elukey: just replied to the patch. Hopefully not a major blocker [08:44:23] arturo: thanks! I suggested another alternative, like using a "if labs -> latest otherwise -> present" [08:44:30] so we'd avoid a flag [08:44:37] but anything is ok for me [08:44:46] that would work too, indeed. Not sure which is more elegant [08:44:55] I'm fine with whatever makes our PCC noop [08:45:37] I'd go for the if + some documentation so it is clear why we are doing it, if in the future anybody will work on it [08:46:16] moritzm: better to have a specific flag or if labs/prod in your opinion? [08:46:28] after this I'll stop harassing people I promise :D [08:47:38] it's ok elukey, just do it! :-P what I should really do some day $SOON is to rebuild the bastions as buster [08:50:11] checking on the realm is frowned upon, I think we even have a CI check for this? [08:50:59] I think it depends on the use cases, but ok I'll go for the flag [09:09:24] arturo: merged, thanks a lot! [09:09:35] 🎉 [09:20:51] I applied the user slice limits to all the analytics clients, I hope that this will stop OOM causing alerts and pain to SRE [09:20:54] we'll see :) [09:41:54] <_joe_> elukey: it will generate new pains I'm sure, but congrats, using cgroups appropriately in systemd is a nice thing indeed [09:50:23] _joe_ credits to arturo for introducing me to the concept + writing docs :) [09:51:16] :-P [09:51:41] I am also interested in expanding the limtis to cpu usage, but I'll leave the memory ones to boil alone for some days to see if any issue arise [09:51:57] or if they need to be tuned [11:07:28] volans: I recall we had issues with icinga downtimw [11:07:33] so I downtime seperately [11:07:48] that's a separate thing [11:08:00] you don't prevent it this way [11:08:07] also, are those new hosts or existing ones? [11:08:12] existing ones [11:09:13] and you are not doing depool or apache fast test? [11:10:09] -a does tha fast test right ? [11:10:38] oh damn I forgot to depool this set, the first set was depooled yes [11:10:40] right, missed that one sorry [11:11:24] I haven't launched the reimage on mw2290 yet btw, happy to try with new options tho [11:13:46] so to re-iterated on the no-downtime, that is just avoiding the downtime before the reimage, and you normally want that, so no reason to use that option [11:14:17] the downtime that sometimes can have issues because of race conditions is the downtime *after* d-i when we run the first puppet run [11:14:29] and that's not tunable because we want that downtime [11:17:37] ok, trying on mw2290 see what happens [11:23:10] effie: mhh the partitioning failed, I'm looking [11:23:21] clearly my partman-fu isn't good [11:23:26] oh [11:23:35] that will fail my reimaging [11:23:42] should we revert for now? [11:24:16] I'll investigate 10 minutes and we can revert yes [11:24:35] we can also move the new partman recipe to a separate file for now and only assign it to m2290? [11:24:53] that makes more sense [11:24:57] good idea [11:25:07] ok I'll do that [11:29:31] effie: reverted, should be good to go [11:29:48] thank you! [13:14:32] hi all i have just pushed https://gerrit.wikimedia.org/r/c/operations/puppet/+/510613 which now means files with a python3 shebang will be tested using python3 flake8 rules [13:15:13] voting or nonvoting? [13:16:30] jbond42: [13:17:13] it runs as the part of the standard ci checks which means its voting like the pytyhon2 flake8 check [13:20:16] i have also created (https://phabricator.wikimedia.org/T239334) to log any complaints and descuss changes to the default style rules [13:22:55] subscribed, thanks [13:24:36] np [13:24:52] jbond42: there seems to be some breakage for files containing neither Py2 or Py3? https://integration.wikimedia.org/ci/job/operations-puppet-tests-stretch-docker/27070/console [13:24:54] NameError: undefined local variable or method `python2_files' for # [13:25:08] for a puppet patch (https://gerrit.wikimedia.org/r/#/c/operations/puppet/+/553331/) [13:25:23] ok thanks ill take a look [13:26:58] ack, thx [13:29:41] sending a fix now [13:34:40] moritz fix in place and i have rebased your change [13:35:12] thanks [13:49:07] Hi moritzm [13:51:15] I'm also having the same NameError problem with my puppet patch https://gerrit.wikimedia.org/r/c/operations/puppet/+/552107 [13:51:37] https://integration.wikimedia.org/ci/job/operations-puppet-tests-stretch-docker/27071/console [13:52:07] phamhi: you should be able tio fix it via a rebase [13:53:23] yeah, rebase will fix it [13:53:40] let me try it out..thanks [13:54:25] it worked..thanks [13:55:21] that change is quite a nice test for the actual changesi made so thanks :) [13:55:42] haha :)