[00:06:21] * ebernhardson finds it a little amusing the logback fix was 48 changed files  with 35 additions and 3,682 deletions. They basically deleted a few thousand lines of code :)
[00:06:54] <ebernhardson>	 probably a good call, drop the functionality all together
[00:49:05] <ryankemper>	 ebernhardson: could use help figuring out where we actually specify the logback version for query service
[08:09:49] <dcausse>	 wcqs-beta import eta is 400 over 1435 files to import
[09:27:29] <gehel>	 mpham: would you have a bit of time to discuss yesterday's meeting?
[09:27:44] <gehel>	 meet.google.com/jcd-dqmc-vrc
[10:40:28] <gehel>	 errand + lunch
[11:46:42] <dcausse>	 lunch
[13:38:15] <gehel>	 dcausse: we're looking at T214680 with mpham. It seems to me that IDs should stay opaque and subject to change at any point in time. But I also don't understand the impact here. Any chance you have an opinion?
[13:38:15] <stashbot>	 T214680: Document statement URI format for RDF - https://phabricator.wikimedia.org/T214680
[13:39:41] <dcausse>	 looking
[13:58:35] <inflatador>	 Greetings!
[13:58:43] <gehel>	 inflatador: o/
[14:04:38] <dcausse>	 o/
[14:04:43] <dcausse>	 gehel: sorry was distracted
[14:05:55] <dcausse>	 so my take is: the way the statement id are represented in RDF are not owned by us but by the wikibase -> RDF export
[14:06:34] <dcausse>	 for us it's opaque
[14:07:22] <dcausse>	 if the change is just adding a "precision" (hint in the doc) on how this IRI is formed I have no objection
[14:07:47] <dcausse>	 if it's asking to change how this generated today then this is bigger problem
[14:10:11] <gehel>	 dcausse: ok, so let's ignore this for the moment as not our problem
[14:10:31] <gehel>	 ryankemper: (for when around) can you invite Brian to the bi-weekly SRE meeting on Mondays?
[14:42:23] <elukey>	 dcausse: o/
[14:42:28] <dcausse>	 o/
[14:42:32] <elukey>	 I have shutdown kafka-main2001 some mins ago
[14:42:39] <elukey>	 lemme know if Flink doesn't like it again
[14:42:42] <dcausse>	 sure
[15:23:20] <inflatador>	 gehel or anyone else, if I want to have a personal (but visible) repository for notes, meeting agendas, etc what is the appropriate site for that? people.wikimedia.org ?
[15:23:40] <gehel>	 that would be a good place
[15:23:51] <gehel>	 I do have a few things on github, but not up to date
[15:24:02] <gehel>	 or your personal page on wikitech
[15:24:36] <inflatador>	 That works. FWiW, I've used Confluence for this in the past
[15:26:19] <inflatador>	 Prepare to be impressed! https://wikitech.wikimedia.org/wiki/User:Bking
[15:26:28] <gehel>	 \o/
[15:27:02] <Trey314159>	 inflatador: I use my user page on mediawiki: https://www.mediawiki.org/wiki/User:TJones_(WMF)/Notes  
[15:28:18] <inflatador>	 Nice, I'll just steal this wholesale ;P
[15:58:46] <ryankemper>	 Hello!
[15:59:31] <dcausse>	 o/
[15:59:37] <mpham>	 won't make retro because my 1;1 with shari is right now
[16:02:07] <gehel>	 mpham: we'll survive without you...
[16:02:12] <gehel>	 for everyone else: https://meet.google.com/ssh-zegc-cyw
[16:02:44] <gehel>	 dcausse, ebernhardson, ejoseph ^
[16:02:49] <dcausse>	 oops
[16:03:11] <ebernhardson>	 sec
[16:10:26] <Trey314159>	 mpham, we are going for an hour today, so feel free to join late!
[17:03:29] <ebernhardson>	 ryankemper: i don't know if it's worth trying to do anything about, but i suspect our rolling restarts are what cause failures inside our weekly dumps. Maybe we move the dumps to friday so they are less likely to hit a rolling restart?
[17:03:41] <ebernhardson>	 re: https://phabricator.wikimedia.org/T265056
[17:04:13] <ryankemper>	 ebernhardson: that's my favorite kind of fix, quick and easy :) let's do it
[17:04:26] <ebernhardson>	 ok i'll work something up
[18:11:26] <dcausse>	 dinner
[18:48:24] <inflatador>	 lunch
[19:29:42] <inflatador>	 aaand back
[19:49:49] * ebernhardson is finally getting around to the interview training course. it's lots of fun :)
[20:31:14] <ryankemper>	 ebernhardson: I should probably ask data-engineering this, but do you know which of https://wikitech.wikimedia.org/wiki/Analytics/Data_access#Generic_data_access_(can_go_together_with_the_Team_specific_ones) brian should request? He definitely needs to be in `analytics-search-users` but wondering about the `analytics-privatedata-users` - specifically whether we should request `kerberos` too
[20:33:51] <ebernhardson>	 hmm, hard to say what really matters. analytics-search-users allows to ssh into a few machines and sudo to analytics-search user, i expect can do that already? I guess can't hurt to set it
[20:35:24] <ebernhardson>	 analytics-privatedata-users should grant access inside hadoop which is probably needed in addition to standard ops privs. Looking what the kerberos one does, i don't remember that
[20:36:12] <ryankemper>	 I seem to remember not having kerberos access when wanting to look at some sort of analytics data once...but that was the only time it ever came up
[20:36:30] <ryankemper>	 Good point on search-users being redundant, so probably just the vanilla `analytics-privatedata-users` is sufficient
[20:37:13] <ebernhardson>	 based on the analytics-privatedata-users description in admin/data/data.yaml i expect he does need a kerberos principal
[20:38:07] <ebernhardson>	 if it's used or not depends, i use the analytics-search principal for anything official and my own principal for debugging/testing/etc.
[20:43:07] <ryankemper>	 Okay that makes sense
[20:58:51] <ebernhardson>	 i wonder about the table in https://wikitech.wikimedia.org/wiki/Analytics/Data_access#Data_access_granted though,  the analytics-search-users group grants access to the analytics-search user, which has private data access.  But the table marks it as not being provided, probably because it's not directly provided
[20:58:55] <ebernhardson>	 not our problem though, i guess :P
[21:11:51] <ryankemper>	 heh, that is an interesting oversight
[23:16:09] <inflatador>	 Re: My Onboarding document. Looks like the slack channel "#technology-department" is archived, guessing this has been replaced by #platform-engineering-team ?
[23:45:47] <ebernhardson>	 inflatador: hmm, i have #tech-department
[23:49:21] <ebernhardson>	 platform-eng would be specific to that team, #tech-department should be relatively quiet and i think is a second avenue for anouncements (also, usually, email)