[00:59:00] Amir1: do we have a probable cause for the big reduction in replica queries last few weeks? [00:59:01] https://grafana.wikimedia.org/d/000000278/mysql-aggregated?orgId=1&from=now-90d&to=now&timezone=utc&var-site=codfw&var-group=core&var-shard=s1&var-shard=s2&var-role=slave [00:59:43] https://usercontent.irccloud-cdn.com/file/xcjRd9CP/Screenshot%202026-08-18%20at%2017.59.27.png [01:00:33] I'm looking for impact of T297300 but doesn't line up (went out with 1.47.0-wmf.14 on 4-6 Aug) [01:00:34] T297300: PageProps doesn't cache result when it is false - https://phabricator.wikimedia.org/T297300 [01:01:29] ah no, I was fooled by Gerrit not showing older branches/tags [01:01:53] https://www.mediawiki.org/wiki/MediaWiki_1.47/wmf.13 that's 28-30 July [01:02:13] that lines up but not sure if other stuff went out to take credit instead [03:47:01] * Krinkle is thinking of adding a TransactionProfiler mode for logged-out parser cache hit page views [03:47:18] It'd have a super low limit like 10 or 20 queries in prod [03:51:53] ref T347123 up from 19 to 23... [03:51:53] T347123: Reduce database queries on parsercached logged-out page views (Sep 2023) - https://phabricator.wikimedia.org/T347123 [10:53:32] No idea :( [14:36:09] ok well my change definitely didn't bring that down beause apparenlety it didn't work [23:17:56] The selector panel on this dash suggests that 10% of our flamegraph samples are for the unknown_unknown entrypoint. https://grafana-rw.wikimedia.org/d/78144bd1-5e43-46f9-a5fd-8a6527da0c6f/mediawiki-entrypoint-profiling3a-one-component [23:18:06] that seems like a big chunk to not attribute [23:18:14] note that load_php is 1% [23:18:58] looks like they're all POST requests and spend more of their time in core_JobQueue [23:20:19] https://codesearch.wmcloud.org/deployed/?q=getHandlerMetricPrefix [23:20:39] `ProfilingContext::singleton()->init( MW_ENTRY_POINT, $actionName );` [23:21:00] https://gerrit.wikimedia.org/g/operations/mediawiki-config/+/eb96da83e13f3850565e47cfa20e74fa298319b3/rpc/RunSingleJob.php#1 [23:21:09] Doesn't set MW_ENTRY_POINT I guess :) [23:43:18] ref T435422 [23:43:19] T435422: 10% of flame graph samples attributed to unknown_unknown - https://phabricator.wikimedia.org/T435422