[14:46:42] jynus: Hi, due to T410589 you will get a pretty large reduction in backup sizes triggering alerts. the next one will be coming in 10 hours when eqiad backup of s6 starts. I think it'd be around even 50% reduction unless I'm reading the data wrong [14:46:42] T410589: Optimize all core tables, late 2025 - https://phabricator.wikimedia.org/T410589 [14:47:19] so heads up :D [14:47:50] https://phabricator.wikimedia.org/T410401#11391870 [14:47:55] Nice, so I assume only on raw ones, right? [14:48:44] Semi-related- probably some new tables are not in compressed format, because the conversion was only done once [14:48:58] new ones may have been created, accidentaly reverted on recovery, etc. [14:49:13] yeah, not logical backups [14:49:20] as one thing you may want to check for your TODO (in case it wasn' done in this run) [14:49:41] if they are small, s5 wikis is not a big deal [14:49:45] logical backups might go down a bit in s3 and s7 due to clean up of ruwikinews and viwiki [14:50:07] but if there are any big tables with missing compression, it would be another thing to check [14:50:25] yeah, I can write a script to check [14:50:36] Should I write a ticket with low or something [14:50:36] would you mind creating a ticket and assign it to me? [14:50:38] ah [14:50:46] Great minds think alike xD [14:50:48] genious minds think equally :-D [14:50:49] lol [14:50:59] lol indeed [14:51:00] I swear I wrote it before you typed it [14:52:06] please send a screenshot of this convo to kwakuofori, to proof we get on well! [14:52:50] xD [14:53:00] we even have direct brain connection [14:53:08] bellow T410401 or some other or standalone? [14:53:08] T410401: Optimize all the things (=MySQL tables) - https://phabricator.wikimedia.org/T410401 [14:53:30] standalone is better [15:05:24] https://phabricator.wikimedia.org/T410747 [15:07:56] I think it is important however [15:08:06] :) [15:08:13] because due to the work at T410589 [15:08:13] T410589: Optimize all core tables, late 2025 - https://phabricator.wikimedia.org/T410589 [15:08:44] it could avoid duplicating work- optimizing once and compressing again, rather than on a single run [15:24:38] have a nice weekend [23:42:50] > s6 eqiad snapshot very_wrong_size 2 hours ago 367.0 GB -20.8 % The previous backup had a size of 463.2 GB, a change larger than 15.0%. [23:43:00] let's see how fast it grows back