[08:14:09] <gehel>	 Hello! I was pinged on T421912, but I'm sadly not on this team anymore... pfischer : could you have a look?
[08:14:10] <stashbot>	 T421912: wbsearchentities: strictlanguage parameter does not exclude items with labels only in other languages - https://phabricator.wikimedia.org/T421912
[13:03:14] <inflatador>	 <o/
[13:05:30] <dcausse>	 o/
[13:07:50] <inflatador>	 huge shout-out to brouberol for getting certificate monitoring in place for the opensearch k8s clusters. Check `opensearch_k8s_master_cert_expiry_seconds{namespace="opensearch-semantic-search"}` for an example
[13:11:51] <dcausse>	 brouberol: thanks! :)
[13:13:48] <brouberol>	 welcome :)
[13:17:11] <ebernhardson>	 \o
[13:17:43] <ebernhardson>	 inflatador: sorry didn't responnd,  cleanup is in cirrus as part of the index creation script.  If it fails to create properly it deletes itself.
[13:17:52] <dcausse>	 o/
[13:17:58] <ebernhardson>	 there shouldn't be any reindexes running anyhwere
[13:19:41] <inflatador>	 no worries, I would've done it yesterday anyway but we got distracted on wdqs reimages
[13:56:57] <dcausse>	 just realized that now the completion suggester affects the score based on the length of the match before the first typo
[13:57:04] <dcausse>	 not sure when they implemented this
[13:58:40] <dcausse>	 the score seems to be nb_prefix_char * doc_score which means, if the typo is near the end of a long query the fuzzy match could be massively boosted
[13:58:48] <ebernhardson>	 :S
[14:16:12] <inflatador>	  just a heads-up that I'm deploying a software firewall change  ( https://gerrit.wikimedia.org/r/c/operations/puppet/+/1266372 ) should be a no-op but just FYI
[14:31:54] <moritzm>	 if you want to roll this out extra carefully you can also:
[14:32:02] <moritzm>	 1. disable Puppet for the role
[14:32:13] <moritzm>	 2. dump iptables -L
[14:32:17] <moritzm>	 3. apply on a single node
[14:32:38] <moritzm>	 4. dump iptables -L and compare with the previous state to validate it's a NOP
[14:32:57] <moritzm>	 that's why I usually do when rolling out conversions from ferm::service to firewall::service
[14:34:08] <inflatador>	 Thanks moritzm , I did cloudelastic -> codfw -> eqiad just now, looks like everything worked as advertised ;)
[14:35:15] <moritzm>	 ok :-)
[14:35:21] <inflatador>	 I was just polling the opensearch API, if the cluster ports are blocked they'll drop out of the cluster pretty quickly
[15:23:38] <ebernhardson>	 somehow it never ends...cindy is now having problems with an expired deb@sury.org key :P  updating...
[15:26:57] <dcausse>	 :/
[15:35:26] <ebernhardson>	 curiously, it doesn't have that issue locally using the same images :S
[16:40:42] * ebernhardson needs a parallelized git pickaxe...it's so slow against mw core
[16:50:45] <ebernhardson>	 huh, SearchPostgresql  does a die('Sorry, ...') if it's initial sql query fails
[16:51:56] <dcausse>	 wow
[17:16:50] <dcausse>	 dinner
[21:38:24] <inflatador>	 how do y'all interact with YARN usually? Ben is out and I'm trying to help, just wondering if y'all login to a stat host, use the web UI or something else