[10:18:39] <dcausse>	 lunch
[12:16:36] <pfischer>	 ebernhardson: Looks like the IP range check for the NetworkSession extension is optional, see T357353. Should we keep it nonetheless and simply configure it to pass any source IP?
[12:16:37] <stashbot>	 T357353: Application Security Review Request : NetworkSession  MediaWiki extension  - https://phabricator.wikimedia.org/T357353
[12:48:18] <dcausse>	 image_suggestions_weekly finally happy and was able to skip empty partitions
[12:58:20] <inflatador>	 <o/
[12:59:11] <dcausse>	 o/
[13:19:59] <gehel>	 Reminder to fill in the team health check before Monday (see email)
[14:04:44] <ebernhardson>	 meh, took a quick look over the reindexing progress. codfw finished commonswiki and is working through things. eqiad failed the reindex 1106 rejecting a request due to filling it's thread pool.  We need to figure out whats occasionally blowing those up :(
[14:05:02] <ebernhardson>	 it starts over, no intervention, but adds a day :P
[14:11:55] <dcausse>	 the "no intervention" part is great :)
[15:03:55] <dr0ptp4kt>	 pfischer dcausse should i schedule pairing time for 1600 / 4 PM your time tomorrow or would monday around that time maybe be a better slot?
[15:04:40] <dcausse>	 dr0ptp4kt: when you want, tomorrow is fine, monday too
[15:05:04] <dr0ptp4kt>	 thx dcausse . pfischer preference?
[15:10:31] <pfischer>	 Tomorrow works for me.
[15:15:24] <pfischer>	 dcausse: Thank you for the image_suggestions_weekly fixes!
[15:15:40] <dcausse>	 np!
[15:22:32] <dcausse>	 ebernhardson: in cirrus-reindex-orchestrator you never need to infer the project hostname from the wikidb?
[15:27:24] <dcausse>	 can do https://en.wikipedia.org/w/api.php?action=query&meta=siteinfo to infer the wikiid from the hostname but was wondering what's the proper way to infer en.wikipedia.org from enwiki
[15:28:27] <ebernhardson>	 dcausse: i've previously used hacks to invoke eval.php and echo php variables. There is also noc
[15:29:02] <ebernhardson>	 dcausse: you can query https://noc.wikimedia.org/wiki.php
[15:29:33] <ebernhardson>	 that script is docroot/noc/wiki.php in mediawiki-config
[15:30:14] <dcausse>	 ah indeed, https://noc.wikimedia.org/wiki.php?wiki=enwiki&format=json might not be too hard to parse
[15:30:23] <dcausse>	 thx!
[15:33:24] <ebernhardson>	 dcausse: one potential issue i just noticed, apparently all unknown wikis are enwiki
[15:33:52] <ebernhardson>	 intentionally: $selected = in_array( $selected, $wikis ) ? $selected : 'enwiki';
[15:43:29] <dcausse>	 ouch
[15:45:30] <dcausse>	 sigh... the noc output does not even have the wikiid do check that the answered config matches the wikiid requested
[15:46:29] <dcausse>	 might have to hardcode this behaviour in my client...
[15:47:26] <inflatador>	 workout, back in ~40
[15:48:47] <ebernhardson>	 dcausse: you could add a strict=true parameter to the script and have it error out
[15:48:55] <ebernhardson>	 i imagine as long as its non-breaking no one would mind
[15:49:30] <dcausse>	 ah sure, will try this
[15:53:37] <dcausse>	 if only I find where the source code is hosted :P
[15:55:25] <ebernhardson>	 dcausse: mediawiki-config/docroot/noc/wiki.php
[15:55:39] <dcausse>	 ah of course :)
[15:58:09] <dr0ptp4kt>	 dcausse pfischer i'll actually set it to start 30 minutes later tomorrow. i'm going to the cowork space and then to meet the local folks afterward, so slight change in schedule for me
[16:01:04] <pfischer>	 dr0ptp4kt: alright, I might not have a full uninterrupted hour then, but still works for me.
[17:07:08] <inflatador>	 back
[17:10:51] * ebernhardson forgot how complex the cirrus index creation is...so many validator classes :P
[17:11:39] <ebernhardson>	 but indeed it's supposed to be able to compare the index configuration against expected and only recreate if necessary, never seen it actually do that :P
[17:13:19] <dcausse>	 yes... I think I've seen a few times a long time ago, it's very brittle...
[17:14:33] <pfischer>	 dinner
[17:14:43] <dcausse>	 or perhaps use justMapping as a first try and if it works skip to the next index?
[17:14:59] <dcausse>	 ah no
[17:15:25] <dcausse>	 that won't work, it won't attempt a change of the analysis settings
[17:15:37] <dcausse>	 dinner
[17:22:09] <ebernhardson>	 hmm, actually because we run with --indexIdentifier=now it will always think that an index needs to be created. In that case it doesn't ever compare the old and new index afaict
[17:22:31] * ebernhardson suspects that parameter is at the wrong level of abstraction
[20:12:08] <inflatador>	 ryankemper re: pairing it looks like David's query might work for alerts https://phabricator.wikimedia.org/T331405#8806909
[20:13:47] <inflatador>	 we were working on T361114 ^^
[20:13:48] <stashbot>	 T361114: Alert Search Platform and/or DPE SRE when Wikidata is lagged - https://phabricator.wikimedia.org/T361114
[20:17:57] <inflatador>	 something like `rate(org_wikidata_query_rdf_blazegraph_filters_QueryEventSenderFilter_event_sender_filter_StartedQueries{cluster="wdqs-internal"}[5m])` 
[20:19:04] <inflatador>	 I **think** we need to distinguish between when Wikidata is lagged as opposed to WDQS (maybe check internal clusters only?) but not sure. Will ask David tomorrow if he's around
[20:19:17] <inflatador>	 (or if anyone else knows, feel free to jump in ;P )
[20:40:51] <inflatador>	 looks like the graphite-based metrics come from https://gerrit.wikimedia.org/r/plugins/gitiles/analytics/wmde/scripts/+/refs/heads/master/src/wikidata/maxlag.php 
[20:59:34] <gehel>	 inflatador: I can also give you more context if needed (tomorrow)
[21:00:27] <inflatador>	 gehel go to bed ;P