[09:05:18] <Amir1>	 marostegui: the good news is that s2 pk change drops 140GB per relica, the bad news is that it's taking 10 hours, meaning to finish codfw, it'll need 80 more hours
[09:05:44] <marostegui>	 nice one
[09:05:53] <marostegui>	 It doesn't matter, I am not in a rush for s2 codfw switch
[09:06:32] <marostegui>	 I am switching s8 primary eqiad tomorrow morning though
[09:07:40] <Amir1>	 That'd be nice. I'll be able to start dropping stuff from s8
[09:08:10] <marostegui>	 yep, I need to migrate s8 master and run a few schema changes too
[09:32:27] <marostegui>	 https://phabricator.wikimedia.org/T307501#9841193
[09:37:20] <Amir1>	 wohooo
[10:31:58] <Emperor>	 Sigh, disks have started failing in this Ceph cluster and it's still in development
[11:44:02] <marostegui>	 This is such a gotcha https://phabricator.wikimedia.org/T366167
[12:30:21] <Emperor>	 oh, ouch, yes
[16:31:46] <jynus>	 in theory all mariadb backups are not happening on 10.6 hosts
[16:31:53] <jynus>	 s/not/now/
[16:32:22] <jynus>	 but leaving some leftover 10.4 hosts for now, in case we need to revert
[16:32:43] <jynus>	 plus a few other tasks missing (cleanups, decoms, os upgrades)