[08:02:05] Why is wikimedia-sre not listed at https://meta.wikimedia.org/wiki/IRC/Channels? [08:06:45] noone thought about it? :D [08:07:52] yeah, "This page attempts to track and identify the numerous Wikimedia IRC channels" [08:07:56] attempts [08:09:22] I'll add it [08:10:36] +1 [08:11:32] here was me thinking it was our secret club :) [08:36:08] * vgutierrez hides "the good stuff" [08:48:19] that explains why I have to look after swift ;p [11:48:44] Dear SREs who do k8s, I will attempt to deploy a dangerous change in admin, please refrain from doing anything dangerous until we are done, thank you ! [12:14:51] oof, I wanted to go to the zoo feeding tigers bare handed. Well, I'll postpone it... [12:34:22] :D [12:37:33] Emperor: o/ [12:37:58] would it be ok for you if I upgrade the swift eqiad's envoys to PKI> [12:37:59] ? [12:38:31] (one host at the time, depooling / deploying repooling every time) [12:38:42] change is https://gerrit.wikimedia.org/r/c/operations/puppet/+/1028859 [12:39:11] also for oncallers (moving the eqiad swift proxies to CFSSL/PKI, ms-fe1009 was already done yesterday) [12:41:46] elukey: sure (I start being in meetings at 14:00 UTC but am good 'til then) [12:42:42] Emperor: okok I'll do it now then! Do you prefer to review the puppet change or shall I proceed? [12:44:54] fabfur: we support feeding animals, you can go [12:45:22] elukey: I don't think I need to review it if you've a +1 from someone who knows about the PKI stuff :) [12:45:55] okok it was to be sure about the Swift part, perfect [12:46:29] effie: does your request extend to the ML K8ses? [12:46:38] yes please [12:47:00] Alright, please lmk when I can venture into the belly of the beast again :) [12:47:08] sure sure, thank you [12:47:43] (unfortunately I just merged an admin_ng change for ML, sorry if that messes with your changes) [12:48:32] It should be constraiend to _only_ affect the ML k8ses, at least [12:53:40] klausman: by accident? [12:55:00] I +2'd without thinking about Jerkins merging it [12:55:41] οκ [12:55:43] ok [12:56:03] I have not merged my changes yes, so I would appreciate it if you could deploy this now [12:56:10] just to make my diff easier [12:56:19] Will do. [12:56:26] great, let me know when you are done [12:58:22] ash shucks, the change is b0rk. I'll revert my change so you can proceed. [12:58:43] I am proceeding with the swift tls upgrade, one node at the time - ms-fe1010 first [13:00:51] elukey: I am terribly sorry, I had no idea you are working on this [13:01:12] elukey: can you please hold on this change? it broke maps last time [13:01:38] let me pull the task for you [13:02:02] effie: I am already doing it, ms-fe1010 is upgraded (together with 1009 done yesterday) [13:02:26] did we attempted to migrate swift to PKI? [13:02:30] yes [13:02:33] effie: my revert is merged, from my pov you are free to proceed [13:02:39] klausman: thanks [13:03:01] elukey: I am digging for the task [13:03:38] elukey: https://phabricator.wikimedia.org/T344324 [13:03:38] https://phabricator.wikimedia.org/T343987 [13:04:05] effie: ah wait but that is thanos right? [13:04:11] not main swift [13:04:17] ah! [13:04:19] phew [13:04:41] okok :) [13:04:43] ok pebcak, sorry for raising the false alarm [13:06:06] nono please better safe than sorry :) [13:06:11] ms-fe1010 upgraded [13:06:14] and repooled [13:10:50] proceeding with the other nodes [13:15:40] ms-fe1011 done [13:23:16] ms-fe1012 done [13:32:02] ms-fe1013 done [13:32:05] (last one standing) [13:36:19] mw-fe1014 done! [13:36:40] vgutierrez: o/ swift proxies in eqiad are all on PKI, I checked the ATS dashboard and didn't see any horror [13:36:53] elukey: <3 [13:37:13] will do codfw tomorrow! Cc: Emperor [13:37:44] thanks, looking good from here [13:38:41] elukey: I'm starting a bit late tomorrow (ascension day, so choir things); I don't think it's a problem given you've been starting afternoon-UTC, but thought I'd mention it. [13:38:43] elukey: <3 [13:39:32] Emperor: ack thanks! [13:39:34] cdanis: <3 [14:26:39] Dear SRE k8s people, I finished the admin work, you may continue with your stuff [14:28:01] * fabfur has been eaten by a tiger, too late... [14:28:24] lol [14:33:01] effie: ty! [14:57:18] effie: did you perchance not push your change to ml-staging-codfw? I see a diff there but not on the ml-serve clusters. [14:57:59] oh ! I missed the staging one! [14:58:00] (change as in helmfile diff) [14:58:05] no worries, I can do it [14:58:38] and done. [14:58:38] klausman: can you please add this detail in https://wikitech.wikimedia.org/wiki/Kubernetes/Clusters ? [14:58:43] Sure [14:58:59] I was reading the document actually to identify where to deploy my changes [14:59:39] klausman: one last thing, after you deploy this change, you will need to delete the 2 istiod pods in the istio-system namespace [14:59:53] and let them recreate themselves [14:59:54] ack [15:01:48] istiod pods deleted and confirmed recreated