[16:25:10] <bstorm_>	 👋🏻 Hello, jynus.  I just finished updating the views on labsdb1009, which went fine without depool (often works).  Just going to record that in SAL.  1010 and 1011 will probably need depools to proceed with...when that's a good idea.  I'd also started 1012's views updating.  Should I stop that?
[16:25:32] <jynus>	 bstorm_: that went well because 1009 is 100% depooled :-D
[16:25:40] <bstorm_>	 LOL
[16:25:43] <bstorm_>	 Awesome :)
[16:25:46] <jynus>	 1012 should be no problem
[16:25:53] <jynus>	 so there are 2 options
[16:25:55] <bstorm_>	 Good reason not to depool 1010
[16:26:16] <jynus>	 either try to do it witout depooling (I don't think we can work with just 1 host)
[16:26:31] <jynus>	 or wait a bit until maintenance finishes and do it at the same pace than ours
[16:26:36] <jynus>	 or
[16:27:05] <jynus>	 doing a non-full depool, just chaning the weights playing with the existing 10 and 11
[16:27:30] <jynus>	 I am ok with any of those, just you may not have been aware we were on limited resources
[16:27:46] <bstorm_>	 A non-full depool with the weights will probably be sufficient.  I can start with 11 since that has a lower weight now, right?
[16:27:51] <jynus>	 because ongoing compression on 1009: T222978
[16:27:52] <stashbot>	 T222978: Compress and defragment tables on labsdb hosts - https://phabricator.wikimedia.org/T222978
[16:28:01] <jynus>	 bstorm_: indeed
[16:28:01] <bstorm_>	 I see
[16:28:08] <jynus>	 that one is mostly for web
[16:28:15] <jynus>	 so it should be easier to do metadata changes
[16:28:22] <jynus>	 mostly the problem with metadata is on analytics hosts
[16:28:32] <jynus>	 so it looks like a great ide to try it first
[16:28:46] <jynus>	 sadly, the maintenance we are doing (compressing) may take some days per host
[16:28:54] <jynus>	 but it was really needed
[16:28:55] <bstorm_>	 Ok, I'll work on that for now.  I can wait on labsdb1010, since this is just to make sure the old text fields for comment stuff gets removed.  With 1009 out of the pool, 1010 is probably going to be busy in some capacity
[16:29:17] <jynus>	 indeed
[16:29:23] <jynus>	 worse than usual
[16:29:30] <bstorm_>	 Thanks!
[16:29:33] <jynus>	 I can adapt to your needs
[16:29:38] <jynus>	 just give me heads up
[16:30:03] <jynus>	 and maybe we can move analytics to 1011 when all others are done
[16:30:19] <jynus>	 or, pause the maintenance even
[16:35:10] <bstorm_>	 It mostly blocks anomie from working on dropping the actual columns, so he'd know the timetable needed.   
[16:35:54] <bstorm_>	 I'll mention the maintenance going on in the ticket just in case there's more urgency 
[16:42:10] <bstorm_>	 Heh, yeah, it's not going to fly on 1011.  It times out waiting for a lock on one table or another.  I'll hold up the view updates and wait for maintenance to be done on 1009 at least unless I find out it is more urgent.  
[16:51:06] <jynus>	 I think we can do the analytics-web switch when you want
[16:51:17] <jynus>	 I don't see a huge problem with that