[09:31:48] <dcausse>	 errand+lunch
[13:16:14] <inflatador>	 <o/
[13:16:36] <dcausse>	 o/
[13:48:13] <ebernhardson>	 \o
[13:56:33] <dcausse>	 o/
[13:58:08] <inflatador>	 .o/
[14:01:15] <dcausse>	 inflatador: have to do a quick errand to pick up my son, should be back in 20mins for the pairing session
[14:01:55] <inflatador>	 dcausse np, let's go ahead and cancel unless you had anything
[14:08:58] * ebernhardson should really put together a proper container for mjolnir testing...the current version ends up re-downloading 1gb+ of maven packages every tox run
[14:10:01] <inflatador>	 The Maven folks will thank you :) . If I can help with that LMK
[14:10:57] <ebernhardson>	 it's not hard, just work :P Currently we have a plain Dockerfile that can run it, but .gitlab-ci.yml does it's own independant thing. Need to migrate it to blubber so everyone does the same thing
[14:12:07] <ebernhardson>	 i guess it's also weird since python tox builds a jvm package, to then use that package in the test suite
[14:17:26] <inflatador>	 immutable* infrastructure
[14:30:10] <dcausse>	 back
[15:21:54] <ebernhardson>	 google ai is so silly sometimes: query "did dec make vax", answer "No, Digital Equipment Corporation (DEC) did not produce vaccines. DEC was a computer manufacturer, specifically known for its PDP and VAX superminicomputer series,"
[16:19:24] <inflatador>	 LOL
[16:20:00] <inflatador>	 another small patch to fix conftool data https://gerrit.wikimedia.org/r/c/operations/puppet/+/1138406
[16:41:36] <inflatador>	 ryankemper if you're around, are you able to review ^^ ?
[16:46:28] <inflatador>	 looks like cirrussearch2113 is our last non-master to reimage BTW. We should be able to add new masters and roll-restart after that
[17:39:52] <inflatador>	 nm. I self-merged the above. https://gerrit.wikimedia.org/r/c/operations/puppet/+/1137069 is still up for grabs, however ;)
[17:59:03] <inflatador>	 lunch, back in ~40
[18:09:42] * ebernhardson realizes what my lastpass problem is...the url's for everthing changed into a bunch of tofu, which copy/paste says are control characters :S
[18:10:18] <ebernhardson>	 which is probably the url encryption that it turned on, and is now failing to decode
[18:51:10] <ebernhardson>	 Trey314159: for czech diacritics, i wonder if we should reindex eqiad now, before we start migrating hosts, then do it on codfw once we finish the migration and shift all traffic there? Since it's only czech wikis should be reasonably quick
[18:51:15] <inflatador>	 back
[18:51:27] <inflatador>	 also damn, that sucks re: lastpass
[19:13:19] <inflatador>	 do any of y'all have bash functions or scripts for the elastic API? I'm getting sick of typing `curl http://0:9200/_cat/nodes' over and over again. If not I'll whip something up
[19:23:22] <ebernhardson>	 inflatador: nope, i type it out, with the -H 'Content-Type: application/json' all the time too :P
[19:25:31] <inflatador>	 ebernhardson glad to know I'm not the only one ;)
[19:40:44] <Trey314159>	 ebernhardson: I'm not *against* reindexing Czech wikis sooner if it won't cause problems.
[19:44:13] <ebernhardson>	 if we wait for everything, it might be mid may, iiuc we can get it done instead in the next week or two.  I'll kick off the reindex for czech in eqiad later today
[19:44:26] <ebernhardson>	 wikidata/commons will of course wait 
[19:44:55] <ebernhardson>	 i guess eqiad and cloudelastic 
[19:45:14] <ebernhardson>	 which,separately, is a naming conundrum we have decided to ignore :P
[19:46:04] <inflatador>	 'cloudcirrussearch' seems a bit long
[19:58:03] * ebernhardson realizes it was merged last week, but no train this week. reindexing today won't do anything
[20:03:26] <inflatador>	 the amount of formatting changes for elastic vs cirrus in https://gerrit.wikimedia.org/r/c/operations/puppet/+/1138446 are driving me nuts. I guess they weren't consistent with each other? Or I screwed something else up
[20:04:17] <ebernhardson>	 went from 4 spaces to 2
[20:04:52] <ebernhardson>	 inflatador: you can change the setting for showing that, the gear in top right
[20:05:03] <ebernhardson>	 change "ignore whitespace" to "leading & trailing"
[20:13:42] <inflatador>	 ebernhardson ACK, so you think I should fix it to minimize the whitespace changes? Changing that much kinda makes me squeamish
[20:14:23] <ebernhardson>	 inflatador: right answer is probably to be consistent with the rest of the repo. IMO there should be a lint that enforces the "correct" number of spaces in yaml's, i'm not sure what it should be though
[20:14:40] <ebernhardson>	 i guess can do a quick review of files that lots of people touch to get an idea of what should be the standard
[20:15:39] <ebernhardson>	 so like hieradata/common/service.yaml is 2 spaces, same in hieradata/common/kubernetes.  It seems like the old version was incorrect
[20:19:32] <ebernhardson>	 yamlllint finds all kinds of problems :P
[20:22:36] <inflatador>	 ebernhardson thanks for looking, I thinking about that myself
[20:23:20] <ebernhardson>	 according to yamllint, 140 hieradata files fail with indent spaces set to 2, 619 fail if set to 4. Even just setting to 'consistent', which means same within a single file, fails 121
[20:23:36] <ebernhardson>	 which reinforces my idea that an automated CI lint would bring some consistency :)
[20:23:53] <ebernhardson>	 but it does suggest 2 spaces is the expected standard
[20:24:25] <inflatador>	 sounds like the least bad option is to use 2 spaces, then?
[20:24:57] <ebernhardson>	 yea
[20:26:46] <inflatador>	 I'll put a comment to that effect
[20:27:35] <ebernhardson>	 amusingly, yamllint has an option `indent-sequences: whatever`, but it's not really whatever.  It means you can not indent a sequence, or you can indent it properly...but we have places that indent 1 space :P
[20:29:12] <inflatador>	 Yeah,  I think `indices_to_monitor` had 1-space indentation...hopefully I fixed it
[20:29:34] <inflatador>	 yup
[20:29:44] <ebernhardson>	 wonder what it would take to get this actually in ci, i suppose would have to somehow ignore all the existing errors. Allow the same errors, or fewer errors, but don't allow adding new ones
[20:30:29] <inflatador>	 a new ansible repo seems to be the perfect place to start enforcing good habits ;P
[20:32:29] <ebernhardson>	 good luck with that :P
[20:44:43] * inflatador puts in on the pile along with the failed reimages
[20:50:20] <ryankemper>	 hitting traffic otw back with dog, lil late to pairing
[21:01:38] <inflatador>	 ACK
[21:03:12] * ebernhardson ponders how hard it would be to start with https://github.com/antlr/grammars-v4/tree/master/xsd-regex, modify it to match lucene regex syntax, and then do regex->antlr ast->transform start/end markers->stringify
[21:05:39] <ebernhardson>	 surprisingly the language definition isn't huge, about 60 lines (incl whitespace) for the lexer, and 120 for the parser
[21:06:57] <ebernhardson>	 notably though, that definition doesn't support ^ or $ either :P
[21:19:20] <ryankemper>	 inflatador: here
[21:19:38] <inflatador>	 ryankemper ACK,bry
[21:19:40] <inflatador>	 brt
[21:22:42] <ebernhardson>	 i'm suspecting this might not be worthwhile ... it would be nice to support the anchors properly, like `(^|foo)bar`, but maybe we keep it simple and only support ^ as the actual first char, and $ as the last