[02:13:20] many thanks R.aine for taking point on this and K.rinkle for reviewing the proposed updates! also, apologies for missing the mention while focused on T383047 earlier :) [02:13:20] T383047: Could not send confirmation email: Unknown error in PHP's mail() function. - https://phabricator.wikimedia.org/T383047 [06:50:40] I've switched m5 master, which has mailman, if you see/notice something strange let me know [08:01:51] bjensen, marostegui, I'm going to depool drmrs shortly to upgrade the core routers for https://phabricator.wikimedia.org/T431749 [08:06:14] XioNoX: thanks for the heads up [09:21:34] XioNoX: check -operations if you have a sec [09:30:33] repooling drmrs [09:47:32] oh no, T383047 is One Of Those (see last comment) :D [09:47:33] T383047: Could not send confirmation email: Unknown error in PHP's mail() function. - https://phabricator.wikimedia.org/T383047 [13:03:23] cezmunsta: merging your change too [13:04:44] {{done}} [13:12:56] FYI, in advance of maintenance on the conf* (e.g., conftool etcd) cluster in codfw, we're going to start depooling etcd client traffic there (in favor of reading from eqiad) shortly. this is similar to what we did a couple of weeks ago for a rack maintenance, and no impact is expected. [13:12:56] cc: oncallers Raine, federico3 [13:13:10] ack [13:28:26] federico3, Raine, I'm going to depool esams for a router upgrade - https://phabricator.wikimedia.org/T431751 [13:28:37] ack [13:34:24] godog: ty [14:11:03] repooling esams [16:10:57] following up on the conf* work: etcd is back in action on conf2004 post-reimage. investigating some issues that are preventing zookeeper startup along with elukey. [17:07:57] update: I was able to get zookeeper running again with some classpath hackery. I will follow with some puppet patches to persist it, along with other issues discovered. [17:08:18] thanks <3 [17:47:21] anyone feeling particularly interested in reviewing [0] to lock in the classpath hack? as I mention in the comment on the patch [1] [17:47:21] [0] https://gerrit.wikimedia.org/r/c/operations/puppet/+/1318749 [17:47:25] [1] https://gerrit.wikimedia.org/r/c/operations/puppet/+/1318749/comments/2305ae3d_a792a060 [17:47:33] whoops ... hit enter too soon :) [17:48:09] as I mention in the comment on the patch, I suspect that PCC believes this to be a noop for conf2004 due to it only having very recently switched to bookworm. [17:49:02] almost certainly yeah - https://wikitech.wikimedia.org/wiki/Help:Puppet-compiler#Manually_update_production is the incantation if you wanted to wait for it [17:49:46] thanks! I have had ... extremely mixed results with that actually working, so I *probably* won't wait :) [17:50:01] haha valid [17:50:07] (looking) [17:53:19] +1 [17:53:31] thank you! [18:29:16] alright, next one: if anyone might be able to review [0], that would be greatly appreciated. this maintains the status quo in order to get things stable, though I would like to revisit whether we even want these nginx SSL params anymore, given the current state of the certificates we use. [18:29:16] [0] https://gerrit.wikimedia.org/r/c/operations/puppet/+/1318760 [18:54:22] And now one more [1], to fix the v2-API backup timer on etcd v3.4+ - if anyone might be able to review that and [0] above, that would be swell :) [18:54:22] [1] https://gerrit.wikimedia.org/r/c/operations/puppet/+/1318764 [18:58:24] looks good, thanks scott [18:58:54] amazing, thanks cdanis! [19:19:56] alright, I'm going to pause for a bit. I'll likely take up two more follow-up changes today, then call it a day and pick things up early tomorrow. thanks, all, for the reviews! [19:37:21] belated on-call handoff: boring \o/ (despite Sco.tt's best efforts ;-) ) [20:31:34] swfrench-wmf: if you have a moment, I tracked down our email timeout, https://gerrit.wikimedia.org/r/c/operations/puppet/+/1318778 [20:32:18] oh, that's awesome! I'm doing something mildly fiddly, but will take a look soon [20:32:28] thanks [23:01:10] FYI, I'm hands-off on conf* hosts in codfw for the day. current status is summarized in [0] and I expect to pick things back up first thing tomorrow. in the meantime, etcd client traffic remains depooled from codfw. [23:01:10] [0] https://phabricator.wikimedia.org/T428495#12165373