[10:52:59] cleanup-related? "dump for db_inventory at eqiad (db1115, 2022-12-20 03:19:12): 93 KiB is less than 300 KiB" [10:53:41] didn't we drop a couple of tables? [10:53:49] in zarcillio? [10:53:50] I did yes [10:54:00] ok, then lowering the min size [10:54:34] FWIW I have deleted a lot of files in a small set of databases T325050 [10:54:34] T325050: Clean up database hosts with unusual amount of data - https://phabricator.wikimedia.org/T325050 [10:54:46] you might get some alerts for that [10:55:18] db files or non live ones? I will only alert for the first [10:55:34] db2102 (s1) still has directories for s7 wikis in /srv/sqldata, can I drop them? :D [10:56:04] e.g. I don't backup (yet) (bin)logs [10:56:27] jynus: it's different for each host, some have sqldata.bakup directories. Some have extra binlogs in sqldata/, some have have directories for databases that don't belong there anymore [10:57:34] yeah, I won't alert for those, although there isn't also "deleted files monitoring" [10:58:03] let me check db2102 [10:58:16] after I update the alert for db_inventory [11:00:11] I need an easier way to compare backup runs differences, I had thought about comparing directories/databases sizes on a first view, and clicking inside to compare files/tables, that way large dbs like s3 are reasonable to handle, thoughts? [11:00:51] e.g. -> if you go to s3, it will only show the side by size size of databases, and then you click a database to show the side by size comparison of tables [11:01:18] that way we show only max 600 files rather than 200K [11:02:53] Don't make think too close to the holidays :P [11:02:59] he he [11:03:18] I don't think I will work on this this year anyway [11:03:31] it is just that attending these alerts was not trivial [11:03:58] I need an easier way to discover why "backup has reduced by 7%" in size [12:03:43] Amir1: what directories did you meant for db2102 [12:04:24] Cleaned them up already [12:04:55] Stuff like metawiki/ directly under sqldata/ [12:05:10] :-(