[00:00:36] <Teraii>	 ho indeed it's fast :p
[00:01:45] <Teraii>	 thank you a lot ^^
[00:02:51] <Teraii>	 any recommandation for php optimizations ? :)
[00:03:58] <Skizzerz>	 using php-fpm instead of fastcgi will save you some overhead on all of the process startup/teardown stuff but it will be fairly minor
[00:04:07] <Teraii>	 allready done
[00:05:17] <Skizzerz>	 in that case, not much. You already have opcode caching enabled. You can tweak it to stop checking timestamps if this is a production site, but that has its own downsides where it won't recognize right away that underlying files were modified and to reload them if so
[00:05:39] <Teraii>	 the frontal webserver and php-fpm instances are on diffrerent hardware
[00:06:26] <Teraii>	 yea i'm remember tuning the timestamp
[00:06:57] <Teraii>	 but you say something about one process and die before ?
[00:07:11] <Skizzerz>	 yes, but that shouldn't be the case if you already have php-fpm
[00:07:25] <Skizzerz>	 since that was the problem fpm aimed to solve
[00:07:25] <Teraii>	 ha ok
[00:07:36] <Skizzerz>	 do you have multiple backend servers running php?
[00:07:38] <Teraii>	 yes i have a pool of process
[00:07:52] <Teraii>	 yes i have all current version of php
[00:08:02] <Teraii>	 (even php5.5 :p)
[00:08:06] <Teraii>	 php5.6
[00:08:37] <Skizzerz>	 that would likely be the actual reason then. If one php-fpm process serviced the first request and a different php-fpm process serviced the POST to the login form, then that would cause the same issue
[00:08:50] <Skizzerz>	 because apcu is an in-memory cache and not shared between worker processes
[00:09:31] <Skizzerz>	 if it was the same process both times and you were still seeing that, then I have no idea what was going wrong :)
[00:09:34] <Teraii>	 process cannot communicate because all versions of php are in different BSD Jails
[00:10:02] <Teraii>	 even tcp/ip stack are separated 
[00:11:16] <Teraii>	 this wiki running php7.2
[00:12:05] <Teraii>	 so thank you ;)
[16:54:24] <Krinkle>	 addshore: Train blocker / wmf.19 regression - https://phabricator.wikimedia.org/T245062
[17:28:12] <addshore>	 Krinkle: on vacation
[17:28:37] <Krinkle>	 addshore: np, beig taken care of.
[17:28:40] <Krinkle>	 enjoy
[19:52:40] <McJill>	 vacation++
[20:13:36] <quantum>	 Anyone know why Apache 'Header always set Feature-Policy' prevents MediaWiki Tables?  r how to allow them?
[20:20:00] <Vulpix>	 quantum: How does apache prevent tables? can you elaborate?
[20:26:18] <quantum>	 Vulpix: Running a MediaWiki site and use tables to make boxes for side content. (red border around it)
[20:26:55] <quantum>	 When I turn on Feature-Policy in Apache these borders go away.  Text is intact, but no borders.
[20:27:16] <quantum>	 Prob php or js driven.
[20:29:07] <Vulpix>	 What Feature-Policy have you applied?