[09:51:56] <dcausse>	 lunch
[13:14:18] <inflatador>	 <o/
[13:23:13] <ebernhardson>	 \o
[13:25:01] <inflatador>	 dcausse had a question about toolhub_tools index, is it completely gone from our clusters or was it deleted/recreated? Looking at https://phabricator.wikimedia.org/T423993#11854363
[13:25:59] <inflatador>	 Looks like it's in eqiad but not codfw?
[13:28:22] <dcausse>	 o/
[13:28:55] <dcausse>	 inflatador: it's still there indeed, Bryan told me he'll look into upgrading the ones in eqiad after the hackathon
[13:29:12] <dcausse>	 the one in codfw were simply unused so I dropped it
[13:31:38] <inflatador>	 Cool
[15:32:59] <dcausse>	 sigh... there's a bit more dependencies on cirrus_index_without_content than I thought :/
[15:33:08] <ebernhardson>	 :S
[15:34:06] <ebernhardson>	 naively...i guess i thought it only needed the dump?  Or going the other way? Downstream should generally be compatible though
[15:35:07] <dcausse>	 because I'm changing the snapshot format from nodash to dashes I stumbled on things like: https://gitlab.wikimedia.org/repos/structured-data/image-suggestions-data-pipelines/-/merge_requests/12/diffs
[15:36:01] <dcausse>	 but sadly I did all the renames/relocates of the partitions before actually checking this
[15:36:29] <ebernhardson>	 oh, yea :(
[15:46:28] <dcausse>	 tbh I'm still not clear about the drop_old_data scripts... esp why this cirrus_index_without_content did not work well with our custom drop mediawiki snapshots
[15:46:50] <dcausse>	 I guess it's because it has an extra 'cirrus_replica' partition?
[15:48:13] <dcausse>	 anyways, it now works with the generic refinery-drop-older-than script...
[15:48:51] <ebernhardson>	 not sure, it looks like it should work but would probably have to step though the script and see what happens.  Probably would be best if somehow we could get rid of drop_mediawiki_snapshots and have everything run through the same stuff
[15:48:58] <ebernhardson>	 which sounds like what you've done :)
[15:49:08] <ebernhardson>	 well, for this use case
[15:49:51] <dcausse>	 yes... there are still some other tables but going through the renames is a bit of pain...
[15:51:57] * ebernhardson is not really sure what to do with the wikibase tickets...wikibase is as mysterious to me as search is to other people :P
[15:53:19] <dcausse>	 which one?
[16:00:32] <ebernhardson>	 there are two, one about monolingual content and the other about strictlanguage
[16:02:07] <dcausse>	 oh yes... the monolingual one is a bit mysterious...
[16:02:38] <ebernhardson>	 both are things i could technically do, probably, but i don't know enough about whats going on to evaluate if what i did is what they want :P
[16:03:15] <dcausse>	 the strictlanguage one Itamar has a WIP patch (https://gerrit.wikimedia.org/r/c/mediawiki/extensions/WikibaseCirrusSearch/+/1265474) but not sure what they are waiting for, I thought they wanted to metrics to get a sense of the impact
[16:03:38] <ebernhardson>	 yea, it was unclear who that one is waiting on
[16:04:04] <dcausse>	 I can try to ping him tomorrow
[16:04:41] <ebernhardson>	 i suppose i'll just skip those, my next bit for os 2.x was to go around and see if we can remove _doc from everything. I think in 1.x the only valid value for types is _doc, so in theory we should be able to get rid of them all
[16:04:57] <dcausse>	 sure
[16:05:25] <itamarWMDE>	 Yeah, I posted some metrics on the bug ticket itself, I can remove from WIP so that your team can own it as was suggested by Leszek.
[16:06:10] <dcausse>	 itamarWMDE: sounds good, I'll review tomorrow (it might need a quick rebase tho)
[16:08:56] <itamarWMDE>	 cool, I think I will only have time to get to it tomorrow as well, I'll do the rebase then
[16:09:19] <dcausse>	 thanks!
[16:15:36] <dcausse>	 dinner
[18:08:44] * ebernhardson ponders if there is any reason to keep the CheckIndexes maint script