[01:41:45] <a930913>	 Oh ****, I may have just overwritten an important directory without any easy backups available...
[01:42:00] <a930913>	 Please tell me that there are backups? :)
[01:42:46] <a930913>	 Wait, maybe not.
[01:42:54] <a930913>	 But there are backups, right?
[01:47:41] <Krenair>	 Coren, ^
[01:47:51] <Krenair>	 a930913, what directory where?
[01:57:46] <a930913>	 Krenair: Erm, this was in data/project/<bot>
[02:39:55] <Pathoschild>	 Coren: you mentioned batching queries to minimise latency with crosswiki tools (bug #55929), but it looks like PHP explicitly prohibits batching multiple queries together (apparently to prevent statement injection). That makes batching pretty hard for tools that need to fetch a bit of data from each wiki. Did you have something else in mind?
[02:41:28] <Jasper_Deng>	 speaking of which, why has the TS db been down for weeks now?
[02:41:35] <Jasper_Deng>	 (besides the security issue)
[02:41:36] <Earwig>	 Pathoschild: well, PHP is dumb. can you turn that off?
[02:44:16] <Pathoschild>	 Jasper_Deng: it's not down, but there are problems with it. DaB is no longer actively maintaining the Toolserver, so it's starting to fall apart.
[02:44:32] <Pathoschild>	 Earwig: not that I found.
[02:44:47] <Earwig>	 hmm
[02:46:59] <MrZ-man>	 multiple threads? or does PHP suck at that too?
[02:47:40] <Pathoschild>	 I've been doing some research into that today. As far as I can tell, PHP sucks at that too (and big time).
[02:48:25] <Krenair>	 PHP sucks at something?
[02:48:26] <Krenair>	 stop.
[02:48:27] <Krenair>	 the.
[02:48:28] <Krenair>	 presses.
[02:49:15] <Pathoschild>	 Those particular presses were stopped a long time ago.
[02:58:36] <Coren>	 Pathoschild: I mean batching the fetches, or parallelizing small queries.
[02:58:57] <Coren>	 Although, to be fair, I've not tried that with PHP.
[03:04:24] <Pathoschild>	 Coren: PHP isn't very helpful for either of those. What languages were you thinking of?
[03:05:08] <Pathoschild>	 Maybe I'll just port my tools to something else while I'm at it.
[03:11:54] <Coren>	 Well, my weapon of choice tends to be Perl.  But I've seen good batching with the python oursql module.
[03:17:38] <Pathoschild>	 Batching in oursql looks nice.
[03:20:01] <a930913>	 "appears to be non-functional at this time" :|
[03:26:32] <Coren>	 a930913: What does?
[03:29:00] <a930913>	 Coren: The web message on my cgi script.
[03:30:17] <MrZ-man>	 I tried using oursql, but it seems kind of ... broken
[03:30:31] <MrZ-man>	 I was going to file a bug, but it also looks unmaintained
[03:30:44] <Coren>	 I know lots of people use it; it was one of the first requests I got for python packages.
[03:31:05] <Coren>	 a930913: What does you error log say?  That's where you'd find the more interesting information.
[03:32:16] <a930913>	 Coren: What error log?
[03:32:55] <Coren>	 a930913: That depends on a few things; are you using the new webservice scheme, and is your script in PHP?
[03:33:23] <a930913>	 cgi-bin w' python
[03:34:39] <Coren>	 Hm; you'll need to add logging to your bot itself then, or switch to the new scheme (recommended anyways)  https://wikitech.wikimedia.org/wiki/Nova_Resource:Tools/Help/NewWeb
[03:38:28] <MrZ-man>	 If the result of a query wasn't immediately available, fetchone() would return None. Then, if you did another query right after and called fetchone() again, it would give you the result from the first query
[03:38:34] <a930913>	 Coren: -bash: webservice: command not found ?
[03:39:55] <Coren>	 a930913: ?  Where are you getting that error?  /usr/local/bin/webservice is there -- did you override your path to explude /usr/local/bin (it's there by default)
[03:40:23] * Coren  stares at tools-login.

[03:40:33] <Coren>	 People *still* running bots on it.
[03:40:38] * Coren  pulls out the LART.

[03:42:01] <a930913>	 Coren: Oh, I'm on dev?
[03:42:55] <Coren>	 Oh, hah, never thought of putting it on -dev too.  Hang thight.
[03:43:01] <Coren>	 tight*
[03:43:56] <a930913>	 Dw, started on login.
[03:45:19] <Coren>	 a930913: It's on -dev too now.
[03:54:55] <jeremyb>	 Coren: do individual tools typically have pages somewhere? wikitech?
[03:55:54] <Coren>	 jeremyb: Not systematically; I've seen many maintainers put links to some descriptive page from the tools.wmflabs.org landing page though.
[03:56:20] <jeremyb>	 Coren: well, anyway, tools are really cheap, right? :)
[03:56:42] <Coren>	 Yep.
[03:57:29] <jeremyb>	 can i get a nara tool? and add dominic to it?
[03:57:38] <jeremyb>	 he put in a request for tool labs access already
[04:03:02] <jeremyb>	 i'm wondering if we shouldn't thinking about something like lxc
[04:03:28] <jeremyb>	 i want to be able to do stuff like run semi-trusted with a time limit as a less privileged user
[04:03:40] <jeremyb>	 a la math/svg/etc.
[04:03:57] <jeremyb>	 (or at a different level of the stack: docker)
[04:38:11] <jeremyb>	 Coren: is there a better place to put that req? i looked for at least a min for where to request a tool
[04:38:19] * jeremyb  heads to sleep, see you later :)

[04:38:31] <Coren>	 jeremyb: You don't need to request a tool.  Just create one.  :-)
[04:38:39] <jeremyb>	 ohhhh
[04:38:43] <jeremyb>	 huh? :)
[04:38:47] * jeremyb  spins head

[04:38:52] <Coren>	 https://wikitech.wikimedia.org/w/index.php?title=Special:NovaServiceGroup&action=addservicegroup&projectname=tools
[04:38:57] <jeremyb>	 Coren: anyway, approve him? :)
[04:39:17] <Coren>	 I don't have my second factor token with me.  Will do tomorrow.
[04:39:42] <jeremyb>	 aha
[04:42:58] <jeremyb>	 i wonder if we shouldn't have autopatrolled? i see people i know aren't crazy in recent changes
[04:43:28] <Pathoschild>	 Coren: Yep, oursql is great at batching so far. Time to port everything to python. :)
[04:43:53] <jeremyb>	 what's a good way to get all subcats for a cat? can't just use sql right?
[04:44:07] <jeremyb>	 my quick hack was to manually expand them all in the web UI
[04:44:15] <jeremyb>	 and copy/munge the HTML :P
[06:09:28] <Pathoschild>	 Coren: okay, so oursql can't actually do batching either. (You can queue multiple statements, but it executes each one in its own round trip.)
[06:09:54] <Pathoschild>	 So at this point I think I'll just wait for the datacenters to be combined.
[07:23:16] * Jasper_Deng  huggles Pathoschild

[10:02:59] <johang>	 why doesn't the exec instances have qsub and similar tools?
[10:17:32] <petan>	 !petan-build
[10:17:32] <wm-bot>	 make -j `getconf _NPROCESSORS_ONLN` deb-pkg LOCALVERSION=-custom
[13:30:08] <Gnumarcoo>	 mmh, how to get welcome.py on wmf labs?
[16:49:25] <ori-l>	 !log grrrit-wm stopped receiving events. restarted it; didn't help. then restarted gerrit-to-redis, which seems to have fixed it.
[16:49:25] <labs-morebots>	 grrrit-wm is not a valid project.
[16:49:31] <ori-l>	 sigh.
[16:49:40] <ori-l>	 !log gerrit-to-redis grrrit-wm stopped receiving events. restarted it; didn't help. then restarted gerrit-to-redis, which seems to have fixed it.
[16:49:40] <labs-morebots>	 gerrit-to-redis is not a valid project.
[16:49:49] <ori-l>	 !log tools grrrit-wm stopped receiving events. restarted it; didn't help. then restarted gerrit-to-redis, which seems to have fixed it.
[16:49:52] <labs-morebots>	 Logged the message, Master
[16:49:59] <ori-l>	 there we go.
[18:49:17] <a930913>	 Hmm, why would ./script give a different output to the web run version?
[19:37:09] * a930913  pokes Coren.

[21:17:06] <giftpflanze>	 Coren: now that the ts will be shut down do you have any updates on email-adresses reachable from the outside?
[21:17:51] <giftpflanze>	 *addresses meh
[23:47:57] <giftpflanze>	 is there a outgoing net bandwidth limit?