[21:58:21] <volans>	 hey, need help?
[22:07:14] <volans>	 jynus: from where did you got the 26 factor?
[22:09:15] <jynus>	 from tendril
[22:09:29] <jynus>	 https://tendril.wikimedia.org/host/view/db1048.eqiad.wmnet/3306
[22:09:45] <jynus>	 or its master https://tendril.wikimedia.org/host/view/db1043.eqiad.wmnet/3306
[22:12:07] <volans>	 got it :)
[22:24:56] <volans>	 I don't get one thing, if it's slow to replicate we shouldn't see at any point a query on show processlist in the replica thread? I see mostly reading from relay log, is it just slow to read those long lines?
[22:31:56] <jynus>	 you could, but not if they are 0.01 second queries
[22:32:00] <jynus>	 just a lot of them
[22:32:12] <jynus>	 for replication, you can check the binary log
[22:32:26] <jynus>	 lots of events there
[22:32:52] <jynus>	 if it was select, it would be impossible with the current monitoring "deaths by a thousand cuts"
[22:33:13] <jynus>	 hence the deployment of performance schema, which allows monitoring of all queries
[22:33:26] <jynus>	 that is why it is worth a 4% overhead
[22:34:23] <volans>	 :) yeah I saw many insert into worker_taskdata
[22:36:51] <jynus>	 I will show you the greatness of p_s once we finish all pending provisionings
[22:37:49] <volans>	 ok, I need to practice a bit with it, I usually was getting enough info from innotop