[07:35:36] o/ [07:43:57] o/ [07:48:01] dcausse: You mentioned that you were able to run cindy in parallel, have you committed those changes already? I only see https://gitlab.wikimedia.org/repos/search-platform/cirrus-integration-test-runner/-/merge_requests/27 but that has not been touched in a couple of days. [07:51:21] pfischer: indeed, it's still on my local repo [07:51:51] was waiting for the chain of patches from Erik to be scanned by cindy before messing up again with cindy [08:01:59] Oh, missed that. Alright. I also pushed some changes to make cindy run locally on a mac, but I guess that can also wait for Eriks CirrusSearch patches merged. Would be interested to learn if this still works under native linux. [08:08:37] pfischer: sure I'll test this soon [09:02:00] pfischer: I believe some of the changes you made (obsolete sury key and the wrong $MW DOCKER call are already in https://gitlab.wikimedia.org/repos/search-platform/cirrus-integration-test-runner/-/merge_requests/27/diffs#cba2451c135f5a81e87f247558fd118814b1b1ce [09:03:37] Oh, good. I’ll rebase it. Thanks for checking! [09:03:58] to make this more complicated cindy now requires an unmerged change in mw cli (https://gitlab.wikimedia.org/repos/releng/cli/-/merge_requests/651), I compiled it at https://people.wikimedia.org/~dcausse/mw-with-dps-mem-fix (but it might be for intel cpu I'm afraid) [09:31:23] pfischer: pushed a last commit to https://gitlab.wikimedia.org/repos/search-platform/cirrus-integration-test-runner/-/merge_requests/27 & https://gerrit.wikimedia.org/r/c/mediawiki/extensions/CirrusSearch/+/1299433 for the parallel run [09:31:47] dcausse: thanks, looking… [09:32:42] I re-activated my intel macbook, so one thing less to worry about… [09:33:36] yes... I whish we had better ways to ship multi-arch images... [09:58:37] filed T428560 for tracking further optimizations in cindy [09:58:37] T428560: Reconsider how cirrus integration test corpus is built - https://phabricator.wikimedia.org/T428560 [10:17:02] dcausse: hii [10:17:21] translate plugin is working, i'm populating the test index [10:21:09] atsukoito: nice! [10:21:45] i'll make a diff that will connect test and new prod servers, and then we can populate it and disconnect old servers [10:26:55] can we backport the fix to wmf.5, or is it better to wait for the train? [10:29:47] atsukoito: either way is fine if you're in a hurry backporting is totally ok [11:12:37] lunch [12:54:36] \o [12:56:07] ebernhardson: hiya [13:02:39] o/ [13:42:06] o/ [15:06:08] dcausse: if I want to backport those 2 patches, https://gerrit.wikimedia.org/r/c/mediawiki/extensions/Translate/+/1297694 and related, to current branch, do i need to make a copy of it in the current branch? [15:06:20] or is there any tool for it? [15:51:12] atsukoito: you can do cherry-picks in gerrit, there's a button under the three dots top right, here because these patches are dependent you have to cherry-pick both, on the second one you'll have to select the base commit of the first cherry-picked commit [15:51:36] thanks [15:52:55] dcausse: do you have spare time to help me with backports tomorrow morning or afternoon? [15:57:53] atsukoito: sure! [16:23:37] Thanks again for your input in the S & R meeting. I summarised the notes: https://docs.google.com/document/d/1UPaBg6kYoZu0Q7xZ0X_l62pQC2MeQY3HFCipV7x2sOc/edit?tab=t.0#heading=h.472f8skho237 - feel free to comment/edit if you miss anything. I’ll compose a message for #semantic-search later tonight. [16:24:31] dcausse: scheduled, https://schedule-deployment.toolforge.org/window/1781074800 [16:24:58] atsukoito: sounds good, I'll be around [17:00:21] discussing to some folks I'm getting a vague impression that mwcli is going to slowly be un-maintained :/ [17:00:30] :( [17:02:17] looking at the multiwiki in mw-docker it feels rather hackish https://www.mediawiki.org/wiki/MediaWiki-Docker/Configuration_recipes/Wiki_farm [17:03:02] involves rebuilding the image to hack the apache config to do the wiki routing via a path :/ [17:04:45] yea it was much clearer in mwcli. At least it's possible so could be refined