[10:22:50] <_joe_>	 is it ok to list jynus and volans as running scripts/performing checks during the switchover as far as databases are involved?
[10:43:45] <jynus>	 ?
[10:43:56] <jynus>	 I am documenting those as we speak
[10:44:07] <jynus>	 https://wikitech.wikimedia.org/wiki/Switch_Datacenter/Manual_cache_warmup
[10:44:16] <jynus>	 https://wikitech.wikimedia.org/wiki/Switch_Datacenter/db_read-only
[10:57:46] <_joe_>	 jynus: yeah at the last rollout meeting it was said we will make a list of "point persons" who would follow a specific aspect of the rollout
[10:57:55] <_joe_>	 s/rollout/switchover/g
[10:58:11] <_joe_>	 so since no one did it, I am compiling it
[14:16:55] <_joe_>	 jynus: whenever you have time, https://gerrit.wikimedia.org/r/#/c/283974 should be ok; it's a pity I have to merge my changes in two passages
[14:17:15] <jynus>	 it makes no sense for mariadb at all
[14:18:09] <_joe_>	 yeah
[14:18:32] <_joe_>	 also, I cannot submit the puppet change until this is merged, yuck
[14:18:35] <jynus>	 it is copying now, so I actually cannot do much
[14:18:42] <jynus>	 yep
[14:18:48] <_joe_>	 do you think the change is ok?
[14:19:03] <jynus>	 let me check , I do not know if you ammeded my last comment
[14:19:34] <jynus>	 yes, I am not sure about the unless for enabled =fase
[14:19:46] <_joe_>	 you're right, it was onlyif
[14:19:53] <jynus>	 no I mean
[14:19:57] <_joe_>	 my original version was with an if... else
[14:20:00] <jynus>	 it should be different
[14:20:04] <_joe_>	 yeah
[14:20:08] <jynus>	 depending if we want to kill or start
[14:20:12] <_joe_>	 yes
[14:20:17] <jynus>	 maybe you got it, sorry
[14:20:26] <_joe_>	 my original version in ops/puppet was correct
[14:20:36] <_joe_>	 then I had to redo it in the submodule and fucked it up :P
[14:21:30] <jynus>	 wellcome to my day to day, literally: https://media.licdn.com/mpr/mpr/jc/AAEAAQAAAAAAAAK-AAAAJDRlNzZjODgxLWQ5ZWQtNGNjZS05Mjc3LWU1ZjkxNzgxODJjYQ.jpg
[14:21:47] <_joe_>	 I feel it :)
[14:21:55] <_joe_>	 I mean i'm not that far from it
[14:22:30] <jynus>	 actually, the worse part is that I know the thigs I have to do wrong
[14:22:42] <jynus>	 let me see, I think I have time now that it is copyinh
[14:22:48] <jynus>	 so I may not need you
[14:25:46] <_joe_>	 here it is anyways
[14:26:05] <_joe_>	 the change is ready, and I can do the others too, I know what you want
[14:26:21] <_joe_>	 this could also be a tactic to lure me in fixing a few things in puppet :P
[14:26:48] <jynus>	 wait, with that you do not need a $command
[14:26:52] <jynus>	 let me ammend
[14:26:58] <jynus>	 and take it from here
[14:27:10] <_joe_>	 ok
[14:27:28] <jynus>	 the copy will take 3 hours, so actually I have time
[14:27:49] <jynus>	 the actual process will take 15 hours, but not 15 hours of my work
[14:39:28] <_joe_>	 ok
[14:39:51] <_joe_>	 I got diverted by my laptop failing
[14:54:40] <_joe_>	 jynus: you are doing the changes then?
[14:54:51] <_joe_>	 I've seen you posted a WiP patch
[14:55:27] <jynus>	 yes
[14:55:38] <jynus>	 go back to your responsabilities
[14:55:54] <jynus>	 I have to do extra changes to accomodate the new masters
[14:56:17] <jynus>	 and the copy will take 3 hours
[14:56:18] <_joe_>	 ok
[14:56:25] <_joe_>	 let me know if you need any help
[14:56:52] <jynus>	 some reviews will be great, not so technical, but pure puppet/syntax, etc
[14:57:30] <_joe_>	 will do
[15:01:16] <jynus>	 I am doing it quick and dirty, we can move the pseudo-orchestration to hiera later
[15:01:37] <jynus>	 the '$master=true's
[15:02:20] <jynus>	 actually, maybe I can do it based on the active datacenter for the parsercaches
[15:03:36] <_joe_>	 that would be sensible in that case, yes
[15:03:54] <jynus>	 I only do it with a parameter for the others because there are many
[15:04:17] <jynus>	 here it is only 1 master and one slave per shard, on each datacenter