[02:42:36] <tgr>	 legoktm: bd808: how do I manually restart a rename (on a single wiki)? just manually queue a LocalRenameUserJob with eval.php or is there an easier way to do it?
[02:44:01] <legoktm>	 tgr: terbium:/home/legoktm/fixStuckGlobalRename.php ? but I haven't updated that for the new changes to the rename job constructor
[02:45:29] <tgr>	 ooh, nice
[02:45:43] <tgr>	 is there a reason that should not be part of the extension?
[02:50:23] <legoktm>	 uh, I think I had a patch to add it and never fixed it after feedback?
[03:08:06] <tgr>	 https://gerrit.wikimedia.org/r/#/c/193763/ apparently
[03:08:34] <tgr>	 (the new file: search command in gerrit really rocks)
[16:10:25] <ostriches>	 bd808: So remember when I had a VM where apt was trying to fetch i386 packages instead of amd64 ones?
[16:10:30] <ostriches>	 It still happens.
[16:10:41] <ostriches>	 Funny thing: uname -a says I'm x86_64, lol.
[16:15:20] * ostriches hacks around, adds [arch=amd64]
[16:23:57] <Reedy>	 That seems to have appeared more on recent ubuntu stuff
[16:24:02] <Reedy>	 like when you get the mariadb links
[16:24:37] <bd808>	 ostriches: weird stuff
[21:01:52] <TimStarling>	 RFC meeting starting now in #wikimedia-office: Notifications in core
[21:32:34] <ostriches>	 bd808: if we land the log4j socket input patch to logstash, it's easy to iterate on the submission side. We can always nuke bad entries until we get it right :)
[21:33:07] <bd808>	 ostriches: agreed. that's prettty much how we have brought each new thing online
[21:33:40] <ostriches>	 Shove some logs -> delete -> shove some prettier logs -> delete
[21:33:41] <ostriches>	 :p
[21:36:58] <ostriches>	 True fact, gerrit has an "output your logs in json" mode now. I half considered just throwing those at logstash but i could see them being a weird format and needing too much coaxing. Fihured native log4j was easiest.
[21:38:38] <bd808>	 the deletion "just happens" too. logs last for 31 days and then are gone
[23:57:14] <SMalyshev>	 anybody knows if there is a way to inject messages (Message class) for testing purposes? 
[23:57:39] <legoktm>	 RawMessage ?
[23:57:42] <SMalyshev>	 i.e. if I have code that relates on wfMessage() is there a way to inject it somehow so I could test it
[23:58:14] <SMalyshev>	 legoktm: I need to make wfMessage return different stuff... so I wonder if it's possible
[23:58:29] <legoktm>	 oh hmm, you can hack fake messages into MessageCache?
[23:58:39] <SMalyshev>	 alternatively, maybe change code to use soemthing else (more ibjectable) instead of wfMessage, but what?
[23:59:05] <SMalyshev>	 legoktm: aha, MessageCache, thanks, I'll check it out