[09:09:02] errand [10:19:02] lunch [13:10:41] \o [13:14:52] I will be late to the Wednesday meeting. I have a dental thing I have to take care of and the next appointment is the same time as the meeting. Should be quick, though. [13:31:21] cloudelastic IO is looking great this morning, i think we can be relatively certain we found a fix [13:32:25] {◕ ◡ ◕} [13:32:35] o/ [13:32:57] I'm gonna kick off a rolling restart just to force some shards to move around [13:33:03] +1 [13:33:13] after that I'll probably reimage one or two to make sure Puppet's doing all the things [14:06:50] dcausse got anything for pairing? NP if not, I'm just working on the above [14:15:58] inflatador: no, sorry for missing the call, was distracted [14:21:20] dcausse no worries, I don't have anything urgent [14:23:32] I/O looks good ( https://w.wiki/Mmcm ) but we did get a flink alert for cloudelastic [14:36:10] kind of expected if it recovers [17:20:26] dinner [17:45:53] cloudelastic cookbook is hung up, it appears that cumin can't access cloudelastic.wikimedia.org:9243 anymore. I had that happen to me last week, I think there must be some rate-limiting happening or something [17:47:47] test test [17:51:23] oh, that's not it. David/Cathal already figured it out, ref T425300 [17:51:24] T425300: lvs on cloudelastic1012 is misconfigured - https://phabricator.wikimedia.org/T425300 [18:48:24] Cloudelastic cluster restart is done, gonna try reimaging 1012 (since it has a broken LVS config anyway) [18:53:52] gonna try my hand at writing a puppet role for relforge/beta cluster using https://www.thelinuxvault.net/blog/how-to-run-podman-containers-under-systemd-with-quadlet/ . That should allow us to swap OpenSearch versions quickly without reimaging everything [19:25:23] cloudelastic1012 is back up, readahead looks reasonable [19:56:37] Forgot to fix the hieradata before reimaging and the LVS iface is still goofy. One more time! [21:19:46] OK, looks like 2012 is fixed [21:19:52] err...1012 that is