[07:29:51] dcausse: I think I'm mostly good with my diagrams. If you have some time to review (no emergeny): https://gerrit.wikimedia.org/r/c/wikidata/query/rdf/+/841850 [07:30:20] pfischer_: ^ we talked about diagrams, so that might be tangentially interesting to you as well. [07:30:46] gehel: thanks! will take a look [10:23:01] lunch, back later [10:39:09] lunch&errand [12:57:03] o/ [13:18:35] lunch/errands [14:08:49] will be few min late to wed meeting, this morning is parent/teacher conference [14:56:01] ^^ will also be late, pairing with SO [18:11:19] can we stop elastic2052? It's not a big deal, but i have to filter it from logstash since it's quite verbosely complaining [18:21:14] urbanecm: re: creating bclwikiquote two days ago, were there any problems running the cirrus maintenance scripts? I'm seeing a bunch of write errors to the wiki's search index, i can fix them up but wondering what we can do longer term to prevent this happening when new wiki's are created [18:22:39] (or maybe you only scap'd out the patches, i dunno) [18:58:17] ebernhardson: it took significantly longer than it usually does, and i think it tried to create the indexes a couple of time, but no actual error (I'd report it otherwise) [18:59:25] urbanecm: hmm, it should create the indices three times for three different clusters, odd that it took longer but i guess no particular info. I didn't see anything interesting for that time period in logstash either...oh well :S maybe it will be rare [19:00:27] ebernhardson: i don't think I actually ran the script for all the five wikis i created. Do you want to try the other ones? [19:00:40] Or I can do it and report back if it runs longer again. [19:00:56] urbanecm: i asked in part because david created a few missing indices yesterday, then i was reviewing today and noticed that one was missing too so created it just now [19:01:09] all for new wikis [19:01:24] Ah, so if should be all done now? [19:01:48] yea everything should be created now, we have a script that checks expected indices for all wikis vs cluster state and it's now all happy [19:03:09] i'm partially surprised how many pages they created, bclwikiquote is already in the 1000's :) [19:40:02] any plans for a wdqs deploy coming up? If not i'll ship the eu knowledge graph whitelist today i guess [20:00:11] ebernhardson: no plans [21:40:25] how fun, elasticsearch-hadoop doesn't support index patterns, or even comma separated index names, so would have to create a few thousand dataframes to either union together, or push into async to run a controlled number of dataframes in parallel writing to separate partitions. not fun :P [21:41:09] * ebernhardson sticks to silly code that talks to elastic directly :P