[08:13:39] <moritzm>	 FYI, I'll start with cumin2002 in 45 mins (if backups are not complete by then, please let me know)
[09:45:58] <jynus>	 one backup failed, but not because of a restart, but because of T336326. Let me know when you are done with restarts moritzm so I can do some maintenance
[09:45:59] <stashbot>	 T336326: db1225 crashed - https://phabricator.wikimedia.org/T336326
[10:21:35] <jynus>	 marostegui: https://phabricator.wikimedia.org/T336326
[10:22:16] <marostegui>	 what?
[10:22:46] <jynus>	 too much info, I admit it. the summary is: CPU 1 machine check error detected.
[10:23:21] <marostegui>	 should we ping DevOps?
[10:23:23] <marostegui>	 dcops
[10:23:31] <marostegui>	 and get it replaced?
[10:24:04] <jynus>	 we should at least ask for feedback. BIOS update?
[10:24:29] <Amir1>	 FYI, I'm deleting a couple hundred million rows from echo tables in x1 (wikidata). Context: T318523
[10:24:29] <stashbot>	 T318523: Don't send article-linked notifications for bots - https://phabricator.wikimedia.org/T318523
[10:24:31] <jynus>	 although I see it was done: 2023-03-30T15:12:27
[10:32:10] <jynus>	 I asked for help at T336326#8840362
[10:32:11] <stashbot>	 T336326: db1225 crashed - https://phabricator.wikimedia.org/T336326
[10:32:48] <jynus>	 meanwhile I have to find some hosts, becase we have a few down/unreliable for s2 backups
[11:07:51] <jynus>	 it is nice when problems can be fixed so easy, thank you Jaime from the past! https://gerrit.wikimedia.org/r/c/operations/puppet/+/918428
[12:47:44] <marostegui>	 https://phabricator.wikimedia.org/T334650#8840912
[13:14:30] <Amir1>	 marostegui: When do you want me to reimage s4 to bookworm?
[13:14:40] * marostegui runs
[13:14:41] <Amir1>	 😈
[13:15:11] <Amir1>	 with root, I can run commands under your username, I'd the reimage using that
[14:03:40] <urandom>	 ottomata, gmodena: mediawiki-page-content-change-enrichment is kind of wordy, is there a "<name> for short"?
[14:07:42] <gmodena>	 urandom mmm not really. That's the application name and associated helm chart.
[14:08:36] <gmodena>	 urandom where would you need an abbreviation? Is it to name a swift bucket?
[14:09:01] <urandom>	 account name (and by convention, username)
[14:14:55] <gmodena>	 urandom do you have some guidelines / recommendations for naming accounts?
[14:24:08] <gmodena>	 will this account map 1:1 to the mediawiki-page-content-change-enrichment bucket, or can many buckets belong to an account (with distinct ACLs) ?
[14:26:32] <urandom>	 gmodena: well, we discussed that (though perhaps not entirely successfully).  
[14:28:12] <urandom>	 gmodena: let's try again; I assume you're putting together a platform, and that various teams will own jobs that do checkpointing?  If there were a single Swift account, and a single Swift user, would that user and accompany credentials be shared over all jobs?
[14:28:48] <urandom>	 or from the perspective of implementers, would Swift access be abstracted away?
[14:30:36] <urandom>	 I was planning to create an account, with a single corresponding user, out of an abundance of caution since this didn't seem certain
[14:31:11] <urandom>	 my concern would be someone (by accident of course), clobbering the state of someone else
[14:53:42] <gmodena>	 yeah, we would not want that. urandom ack. Ideally we would like to abstract swift access. Would it be possible to have an account with multiple user, with distinct credentials (not shared across jobs) ?
[14:55:01] <gmodena>	 With this said; for the purpose of this specific application I think that one account with one user would work. This application is a special case, in the sense that it is owned by the team that owns the platform (Event Platform).
[14:55:56] <gmodena>	 an alternative name to mediawiki-page-content-change-enrichment would be mediawiki-event-enrichment. Would this be short enough? 
[14:55:57] <urandom>	 gmodena: yes, it's possible, but I'm not sure we have any Puppetization of access lists.  Managing them manually seems like a bad idea
[14:56:38] <urandom>	 gmodena: I don't know that we *need* to shorten it, I was just curious if there was something more colloquial in use
[14:56:51] <gmodena>	 gotcha. Manual management is a no-go IMHO, I'd rather go with one account - one user.
[14:58:16] <urandom>	 gmodena: ok
[14:58:23] <gmodena>	 mediawiki-event-enrichment <- this is the repo name that hosts the application code. It might contain other applications if the futures, but this would all be owned by event platform.
[14:58:50] <urandom>	 gmodena: Ok, we can go with that then
[14:58:57] <gmodena>	 can we still have  mediawiki-page-content-change-enrichment as a bucket name?
[14:59:10] <urandom>	 yeah, you can create all of the buckets you want
[14:59:31] <urandom>	 I think we discussed that for each, you'd want them as DC-specific variants
[14:59:40] <gmodena>	 yep
[15:00:08] <urandom>	 So mediawiki-page-content-change-enrichment-{eqiad,codfw}
[15:00:53] <urandom>	 at least, that was my understanding; AFAIK, each environment was meant to be independent
[15:01:38] <gmodena>	 That's what we agreed upon (at least looking at the phab task).
[15:02:07] <gmodena>	 will our account have the rights to create buckets, or is it something you need to do?
[15:03:31] <urandom>	 you should be able to, but I can create these we've discussed
[15:04:49] <gmodena>	 great. Either way works  for me :)