[10:37:42] <gehel>	 dcausse: would you have a minute to review https://gerrit.wikimedia.org/r/c/operations/puppet/+/889080 ?
[10:37:52] <dcausse>	 sure
[10:37:57] <gehel>	 Previous change introduced an incorrect syntax
[10:38:24] <gehel>	 I've tested this one manually on one of the miscweb servers and it seems to work (and to do what it's supposed to)
[10:43:41] <dcausse>	 gehel: will this prevent listing files in e.g. https://people.wikimedia.org/~dcausse/
[10:43:43] <dcausse>	 ?
[10:44:23] <dcausse>	 sorry
[10:44:28] <gehel>	 nope, this is only on the config for the query services
[10:44:29] <dcausse>	 it's only for wdqs
[10:44:40] <dcausse>	 the miscweb thing got me confused
[10:45:00] <gehel>	 I've actually just merged it (sorry). The currently deployed version was broken and alerting
[10:45:06] <dcausse>	 np
[10:45:16] <dcausse>	 was about to +1
[10:46:18] <dcausse>	 but still not quite clear why miscweb is mentioned in the commit message tho
[10:47:16] <gehel>	 because the static files are deployed on miscweb and served from there
[10:47:32] <dcausse>	 ah ok
[10:48:04] <gehel>	 That was changed quite some time ago, we wanted to decouple the deployment of the UI from the SPARQL endpoint, so that WMDE was more autonomous.
[10:54:43] <gehel>	 lunch
[11:07:32] <dcausse>	 lunch 2
[13:56:40] <gehel>	 Errand, back in 20'
[14:01:52] <inflatador>	 <o/
[14:06:57] <dcausse>	 o/
[16:03:23] <ebernhardson>	 \o
[16:07:03] <inflatador>	 <o/
[16:15:27] <pfischer>	 o/
[16:28:16] <dcausse>	 o/
[16:30:52] <gehel>	 dcausse: we're in https://meet.google.com/skb-junj-yti
[16:31:01] <dcausse>	 joining
[16:39:01] <inflatador>	 dcausse https://gerrit.wikimedia.org/r/c/operations/deployment-charts/+/889172
[17:01:47] <dcausse>	 inflatador: thanks, looking
[17:02:31] * ebernhardson wonders if airflow 2 now has a reasonable way to change task colors...i can never tell the difference between yellow "runing" and dirty-yellow "up_for_retry"
[17:03:02] <ebernhardson>	 and the orange upstream_failed is similar enough to the dirty-yellow up_for_retry that thats not clear either
[17:09:05] <inflatador>	 workout, back in ~40
[17:47:21] <inflatador>	 back
[17:48:26] <bd808>	 gehel: When you have time, https://phabricator.wikimedia.org/T329319#8602419 is hoping for your feedback on the potential impact of search-chi-eqiad use by Toolhub during the DC switch.
[17:58:39] <gehel>	 cc ryankemper, inflatador ^
[17:59:05] <gehel>	 bd808: there should be no impact, I'll comment on the task after dinner 
[17:59:18] <inflatador>	 bd808 we have no planned maintenances for search-chi-eqiad during the switchover
[18:01:18] <bd808>	 Excellent. Everyone is making it easier for me to ignore the DC switch and that makes me happy. The only thing better would be to actually have multi-DC all worked out for Toolhub, but someday™.
[18:08:10] <inflatador>	 {◕ ◡ ◕}
[18:22:37] <inflatador>	 dcausse do you know if we also need to raise limits in the main_app stanza? rations-deployment-charts/blob/master/helmfile.d/services/rdf-streaming-updater/values.yaml#L5
[18:22:44] <inflatador>	 https://github.com/wikimedia/operations-deployment-charts/blob/master/helmfile.d/services/rdf-streaming-updater/values.yaml#L5
[18:35:15] <inflatador>	 lunch, back in time for pairing
[18:50:07] <dcausse>	 inflatador: no only the task manager, that part concerns the jobmanager and it runs fine with 1.6G
[19:02:39] <dcausse>	 dinner
[19:24:29] <inflatador>	 back
[19:24:40] <inflatador>	 thanks d-causse, I promise not to bug you again today ;)
[20:57:47] <inflatador>	 quick errand, back in ~30
[21:15:32] <inflatador>	 back
[21:22:14] <inflatador>	 ebernhardson FYI, we re-imaged an-airflow1005.eqiad.wmnet to bullseye but it's failing, I think because Spark2 isn't available on Bullseye. Is that OK or should we reimage back to Buster?
[21:24:50] <inflatador>	 Pasted error here: https://phabricator.wikimedia.org/P44641
[21:27:37] <inflatador>	 I guess it's airflow package rather than spark. hmm
[21:27:59] <ottomata>	 cc btullis 
[21:29:00] <ottomata>	 inflatador: this was discussed a bit today in #data-engineering-teamin slack, jsut cced and invited you to that channel
[21:30:42] <inflatador>	 ottomata ACK , looking
[21:31:24] <ottomata>	 inflatador:  mostly that ben is working on upgrading to bullsye, and considering not including spark 3
[21:31:36] <ottomata>	 sorry, not including spark 2*
[21:32:10] <ottomata>	 inflatador:  does search's airflow need spark 2, or are yall on spark 3?
[21:32:30] <ebernhardson>	 the new one should be all spark3, unless we run into some issue
[21:32:44] <inflatador>	 Ah thanks, I wasn't sure ;)
[21:33:11] <ebernhardson>	 the reason for moving to new airflow is basically to simplify the spark3 migration
[21:38:30] <inflatador>	 ottomata looking at the slack room, looks like btullis is working from https://phabricator.wikimedia.org/T329363
[22:34:40] <ottomata>	 nice.
[22:34:46] <ottomata>	 okay just wanted to put yall in touch.
[22:34:50] <ottomata>	 sounds like things are going as planned :)
[22:42:13] <inflatador>	 slowly but surely ;)
[22:44:01] <inflatador>	 we're going to reimage back to Buster for the short-term, so no need to rush thru the above ticket
[23:44:03] <ebernhardson>	 dcausse: i started looking into porting HivePartitionRangeSensor to ease writing the rdf query dag, but it looks like wmf_airflow_common.hive.RangeHivePartitionSensor might meet what we need there.  It's actually significantly more complicated than what we implemented, so if it ends up being a pain it would still be easy to move ours over.