[00:00:14] :) [00:00:34] * addshore cant wait for it all to cometogether :) [00:03:56] Ryan_Lane: you could always write me a faultless php framework for wiki api and wikidata also incorperating the db repls ;p [00:04:25] hahaha. good luck on that one ;) [00:05:34] addshore: Why isn't it working on tools? [00:06:13] Coren: my scripts run if I run them straight on an exec instance but not through qsub or jsub [00:06:28] addshore: What is the result when you try? [00:06:35] * addshore is trying to find it [00:07:11] 10:1 you're bumping your head against the allocation limit. What language is this written in? [00:07:17] libgcc_s.so.1 must be installed for pthread_cancel to work [00:07:28] php [00:07:31] Yep. Running against the ulimit. Lemme guess: php? [00:07:34] * Coren nods. [00:07:39] :D [00:07:48] php is teh evil. It never error checks its malloc()s [00:08:16] Best way: try once with -mem 2G, and when it's running check it's actual usage. [00:08:40] it uses about 16mb tops [00:08:55] addshore: Heh. PHP won't even *start* with just 16m [00:09:09] thats what it uses when its running though ;p [00:09:25] *correction 19 [00:09:45] addshore: More precisely it's using PHP overhead + 19M. :-) [00:09:53] Try it with -mem 2G [00:10:19] jsub -mem 2G 'php /data/project/addbot/wikidata/g.php --lang=de' ? [00:10:53] Yep. Then qstat -j jobnumber to look at its vmem and maxvmem [00:11:34] I'm pretty sure you're going to be surprised at how much a glutton php is. [00:11:42] (Though, admitedly, nowhere as bad as mono) [00:12:22] Coren: same error when running it [00:12:30] o_O [00:12:37] Mind if I try? [00:12:40] go for it [00:16:20] * Coren is confused [00:17:32] :D [00:18:16] Ah-ha! [00:18:31] I didn't even realize it! You're overquoting! :-) [00:18:57] You're trying to run a program named 'php /data/project/addbot/wikidata/g.php --lang=de'. Literally! :-) [00:19:13] Remove the quotes and you get it to run. It doesn't work, but for a real reason. :-) [00:19:45] PHP Warning: require(/home/addshore/.secure/.stathat.key): failed to open stream: Permission denied in /data/project/addbot/config/stathat.php on line 2 [00:20:06] hmm, thats because you dont have permission to that file :P [00:20:49] The problem isn't that /I/ don't have permission for the file, but that local-addbot doesn't. :-) [00:21:21] You /are/ running your bot through the tool account, aren't you? :-) [00:21:43] ahhhhh, is that my problem? :P [00:22:15] ... :-) [00:22:24] so, su to local-addbot? ;p [00:22:32] shorthand: become addbot [00:22:35] :D [00:23:04] >.< I need a password to su to addbot? [00:23:21] su always needs a password. You can /sudo/ to it though. [00:23:26] sudo -iu local-addbot [00:23:37] :) [00:23:51] Or, again, you know... 'become addbot' [00:23:57] Which does that. :-) [00:25:20] oh :P you didnt put it in quotes the first time ;p [00:26:40] addshore: That's because you seem to be overly fond of quotes. :-) [00:27:21] :( I cant chown my files to have addbot as the owner :< [00:30:02] I can also chown them for you if you want. [00:30:36] addshore: ^^ [00:38:46] What does it mean that a file has a T bit in the file permissions and only that bit set? [00:40:13] apmon: which project is this? [00:40:28] it usually means that a brick is hung for that gluster volume [00:41:09] maps [00:41:26] it is only a few files in /data/project/repo of maps [00:42:15] apmon: That the file has been broken by gluster. [00:42:15] apmon: Add to the long list of reasons to be happy we are getting rid of gluster. [00:42:37] can I just delete or overwrite those files? [00:42:58] apmon: one sec [00:48:19] Coren: moved on to another error now :P [00:48:31] addshore: Pray tell? [00:49:25] well, something isnt working quite right with my db but its probably my fault ^^ [00:49:54] Probably. :-P [00:50:07] apmon: try now? [00:51:03] cd: project: Transport endpoint is not connected [00:51:38] before I could at least see some files and write to the directory... ;-) [00:54:49] Ryan_Lane: I can access the directory from maps-tiles2, but not maps-tiles1 [00:55:08] but I see the same 0 byte T files on maps-tiles2 as I saw on maps-tiles1 [00:55:12] gimme a sec [00:55:13] ah ok [00:56:18] maps-tiles1 is back again [00:56:27] I guess those files were split-brained before the shrink [00:56:34] you'll likely need to remove them [00:56:59] OK, no problem [00:58:46] thanks for looking into it [00:59:28] yw [00:59:55] committing the shrink on bots-project [01:02:39] * Hazard-SJ says hello and seeks assistance [01:03:05] :O [01:03:08] * Krenair waves [01:05:57] hey Hazard-SJ [01:06:21] Coren: all volume shrunk [01:06:27] except for the stopped one [01:06:29] *ones [01:06:34] and I can't fucking delete those [01:06:41] so, we'll just consider it done [01:06:49] it doesn't hurt to have down'd peers [01:06:56] Ah. [01:07:14] Although, history has proven /that/ statement wrong at regular intervals. :-) [01:07:34] What's the problem, Hazard-SJ? [01:07:50] Do we now call labstore[34] ripe for reaping? :-) [01:08:08] yes [01:08:18] have at em [01:08:30] first, though... [01:08:43] let me kill all the gluster processes [01:08:51] and let's leave that over night [01:09:01] Ah. See if something goes boom. [01:10:39] Although, again from experience, we know that gluster-related things don't go boom, they go a sort of squishy-wet "splorch". [01:11:52] seems to be fine so far [01:12:01] oh. they go "boom". trust me [01:24:28] "Boom" implies some sort of failing solidity; gluster fails like a cardboard box full of overripe tomatoes; it doesn't explode, it just collapses under its own weight and leaves mush all over the place. :-) [01:26:35] :D [01:26:38] true [01:27:24] Krenair: Sorry for the delay. Would it be better if I ran by scripts in the bastion cron or from an instance? I'm not entirely familiar with some of these things :P [01:27:36] definitely not the bastion [01:27:44] other non-bastion instance [01:28:12] ah [01:28:47] Ryan_Lane, Krenair: And in that case, how would I put myself on another instance? I think I remember seeing something saying not to use bastion for such things, as they'd slow down things for everyone. [01:29:19] ssh instance-name [01:30:03] looks like paravoid disabled one of your crons on november 27th for dosing bastion :) [01:40:46] So shouldn't something like ssh bots-bnr3 work? [01:40:54] Krenair: Yes :P [01:41:10] If you're in the bots project, yes [01:41:30] I am, but I got an error [01:41:45] Oops [01:42:04] Wait, logged in to something else :P [01:42:07] error? [01:42:32] (you are in the bots project btw) [01:42:35] I was on a different server [01:42:42] Yes, I'm in the project [01:43:19] I got in [01:56:54] Am I supposed to use /data/project/bot_name or just use my home? [01:58:54] use /data/project storage [02:00:44] legoktm: That's where I'm slightly puzzled - there is only one directory in /data/project as far as I see :/ [02:01:31] errr [02:01:34] which host are you on? [02:01:47] and when was your account created? it might take up to an hour to show up [02:02:57] legoktm: I'm currently exploring on bots-bnr3, so I guess that's why? My account's fairly old by now :P [02:03:16] * legoktm looks [02:03:31] legoktm: Wait [02:03:41] http://dpaste.de/wiqen/raw/ [02:03:46] wfm [02:04:33] That's when I login on bastion itself in WinSCP [02:06:01] And no matter what I ssh I still have (username)@bastion1:~$ ... is that wrong? [02:07:23] idk [02:07:32] im not sure how WinSCP works [02:07:53] https://wikitech.wikimedia.org/wiki/Access#Accessing_public_and_private_instances [02:07:54] legoktm: The (username)@bastion1:~$ is from PuTTY [02:07:54] try that [02:08:17] and the dir only gives me asher's dir [02:08:19] * Hazard-SJ looks [02:08:45] winscp probably can't connect all the way through [02:08:51] there's a good tutorial for windows somewhere [02:08:54] one sec [02:09:31] https://wikitech.wikimedia.org/wiki/User:Wikinaut/Help:Access_to_instances_with_PuTTY_and_WinSCP [02:10:58] :| [02:11:10] https://wikitech.wikimedia.org/wiki/User:Wikinaut/Help:Access_to_instances_with_PuTTY_and_WinSCP#How_to_set_up_WiNSCP_for_tunneling_through_bastion.wmflabs.org_to_your_instance [02:21:30] I'm getting somewhere now [02:22:12] Thanks :D [02:26:35] yw [02:29:53] I can't create a folder, though :/ [02:29:56] Command 'mkdir "hazard-bot"' [02:29:57] failed with return code 1 and error message [02:29:59] mkdir: cannot create directory `hazard-bot': Permission denied. [02:33:45] Ryan_Lane: ^ [02:34:04] it should already exist... [02:34:08] where? in /data/project? [02:34:25] you'll likely need to talk to petan about that [02:34:27] he's asleep, though [02:35:24] yes, in bots-bnr4 [02:36:02] Ryan_Lane: I'll just email him and give up for the night then :) [03:45:25] Coren|Sleep: I can't create a screen as my tool [03:45:25] local-legobot@tools-login:~$ screen [03:45:26] Cannot open your terminal '/dev/pts/2' - please check. [03:46:22] legoktm: That's normal since your tool doesn't own the controlling terminal -- you do. You can start a screen /before/ you become your tool though. [03:46:44] legoktm: But, more importantly, you aren't supposed to run a tool in a detached screen. :-) [03:48:40] Ok :P [07:42:05] !log tools petrb: removed reboot information from motd [07:42:07] Logged the message, Master [08:18:57] [bz] (NEW - created by: Peter Bena, priority: Unprioritized - normal) [Bug 47115] emails must not be delivered to wmf sysadmins mail - https://bugzilla.wikimedia.org/show_bug.cgi?id=47115 [09:36:40] !ping [09:36:40] pong [13:52:11] [bz] (NEW - created by: Peter Bena, priority: High - major) [Bug 45768] console doesn't show proper errors - https://bugzilla.wikimedia.org/show_bug.cgi?id=45768 [14:03:55] ps faux [14:10:52] hi [14:10:54] Oren_Bochman, do you have a labs shell account already? [14:11:01] yes [14:11:07] been in active for a while [14:11:11] What's your on-wiki username? [14:11:16] oren [14:11:40] That's your wiki name or your shell name? (There is constant confusion between the two...) [14:11:47] Also, what's your project for, and what would you like it called? [14:12:08] it is a MediaWiki Moodle integration [14:12:19] and please call it moodle [14:12:46] Would you like shared storage between instances? [14:13:10] I think so [14:13:45] I'll probably have a dev inst production inst and a MW inst [14:14:08] ok. Should be all set. Lemme know if it misbehaves. [14:14:15] one more thing [14:14:42] can you add me to the Tools project [14:14:52] Sure. Just a second... [14:15:12] ok, done. [14:15:27] is there any info about migrating stuff to this project ? [14:15:35] like bots or a scripts [14:16:20] There's a bit of documentation here: http://www.mediawiki.org/wiki/Wikimedia_Labs/Tool_Labs/Help [14:16:45] You'll probably want a tool group created; at the moment Coren is the one to ask about that. [14:17:06] He's in EDT so should be online pretty soon. [14:17:10] well it is enough to get me started [14:17:58] I want to make a new bot to push and pull stuff beteween moodle and media wiki [14:18:20] but I need to set things up first [14:18:23] ttl [14:22:38] Would it be possible to get gnuplot for the tools project? [14:25:17] fwilson: Probably; best to ask Coren when he appears. [14:25:36] Alright, I will [14:27:22] automatic plotting of recent vandalism statistics :) [14:28:23] andrewbogott: I need an instance with LAMP [14:28:49] I'm not sure what I need to choose from the Puppet info [14:29:04] Oren_Bochman: First you'll want to set up a web security group [14:29:47] Oren_Bochman: Here's a guide to setting up a mediawiki instance: https://wikitech.wikimedia.org/wiki/Help:Single_Node_MediaWiki [14:30:02] For LAMP w/out mediawiki you'll want to follow that guide but stop short of doing the MW puppet config [14:30:26] ok [14:30:42] And instead you'll want to add… webserver::php5-mysql I think. [14:30:45] I'll probably be adding a Moodle pupet conf [14:30:45] Just that one option should do it. [14:31:02] Sure, if you make a Moodle conf then it can just include a webserver class. [14:31:07] I thought so too [14:31:27] but I forgot about adding a security group [14:31:41] Yeah, awkwardly you need to do that before you create the instance. [14:32:03] Fortunately instance creation is now speedy, thanks to Ryan_Lanes work last week. [14:32:17] I read about that [14:33:59] MaxSem: I have updated the packages in /data/project/repo of the maps project now [14:34:28] The packaging scripts I used to build them are at https://github.com/apmon/OSM-rendering-stack-deplou/tree/master/wikipedia/debian [14:35:59] I also tried to set them up on maps-tiles1. Unfortunately something went wrong and now puppet doesn't work anymore [14:36:06] Is Coren here/ [14:36:19] In meeting, will be back in 5 [14:36:46] hi [14:39:51] Sorry. Back. [14:40:22] fwilson: gnuplot you say? [14:41:56] Change on 12mediawiki a page Wikimedia Labs/Tool Labs/Notepad was modified, changed by MPelletier (WMF) link https://www.mediawiki.org/w/index.php?diff=672365 edit summary: [+12] +gnuplot-nox [14:42:20] which puppet defs do I need to be able to config my instance with puppet? [14:42:36] is puppetmaster::self enough ? [14:42:59] Oren_Bochman, I'm not sure I understand the question. [14:43:16] If you want to write puppet code, then you want puppetmaster::self [14:43:20] I want to config my moodle instance with puppet [14:43:23] ok [14:43:30] If you just want to configure using puppet, then… everything already does that. [14:43:41] I want to code it [14:43:58] https://wikitech.wikimedia.org/wiki/Help:Self-hosted_puppetmaster [14:44:01] since we might need a bunch of these for different languages [14:45:15] For me the addition of puppet classes via labsconsole didn't work [14:45:32] apmon: How come? [14:45:45] It said the class wasn't found. [14:45:59] Whereas when I added it directly to site.pp it did find it and run through fine [14:46:07] apmon: Ah, yes, well you need to have it present in the operations/puppet git /before/ you add it to your node. :-) [14:46:20] I did [14:47:07] apmon: Oh, I see what you mean. Yeah, adding the class via the labconsole adds it you your /node/ definition, but it still has to be reachable from site.pp (or indirectly, if it is a module). [14:47:54] apmon: The labsconsole basically just adds "include your_class" to your node definition, it still has to be visible to puppet in the general case. [14:47:56] Was that always the case? I thought the above used to work (around March), but I might have forgotten a step that was necessary [14:48:43] That was all I did in site.pp to make it work. Added role::osm:db to include of the default node def [14:48:44] apmon: Well, you might have been using a module; puppets autoloads them. If your class isn't in a module, then it needs to be added to site.pp [14:48:44] apmon, thanks! [14:49:02] It is a module [14:49:29] https://gerrit.wikimedia.org/r/#/c/36222/ is the module in question [14:49:41] apmon: o_O then it should have worked. Odd. You added it through "Manage Puppet Groups" then turned it on in "Configure Instance"? [14:49:51] yes [14:50:04] What is the project/instance? [14:50:10] apmon, I've already applied puppetmaster::self to maps-tiles[12] [14:50:51] I tried it on maps-tiles1 (which now has a broken puppet altogether after a restart) and on map-tiles2 [14:50:57] sorry maps-tiles3 [14:51:02] broken how? [14:51:03] * Coren checks. [14:51:08] which is a new instance I created yesterday [14:52:29] after a node is self-hosted, you configure it by editing site.pp [14:52:31] apmon: https://wikitech.wikimedia.org/w/index.php?title=Special:NovaInstance&action=configure&instanceid=a43c1612-c13e-4aeb-ae0b-b2a86a14b36f&project=maps®ion=pmtpa [14:52:44] Oh, wait, let me try and restart maps-tiles1. Because it was doing something strange, I might have killed puppet, which might explain why it sais it can't connect [14:52:45] Note that you didn't turn on any of the classes. :-) [14:52:57] I turned them off again, as it didn't work [14:53:01] Ah. [14:53:10] Mind if I try it? [14:53:25] yes, go for it. [14:53:39] Oh, wait, it's puppetmaster::self [14:53:39] It should be role::osm:db and role::osm:tileserver that should work [14:53:43] yes [14:54:02] If a node is puppetmaster::self, it no longer gets update from git automatically -- you have to update it yourself! :-) [14:54:34] Well, the module in question isn't in the main repository (only in gerrit so far) which is why I need the self hosted puppet [14:54:46] Ah. Okay. :-) [14:54:54] but it is present in the local repository [14:55:55] apmon, want me to do it or you're trying it yourself? [14:56:19] apmon... It worked. [14:56:27] notice: /Stage[main]/Osm::Db::Setup::Basic/Postgresql::Createuser[osm]/Postgresql::Sqlexec[createuser-osm]/Exec[echo "CREATE USER \"osm\";" | psql postgres]/returns: executed successfully [14:56:34] Coren: where did you try it? [14:56:40] maps-tiles3 [14:56:51] well, that has those in site.pp under the default node [14:56:57] Ah. [14:56:58] which is what worked [14:57:04] * Coren tries it again after removing that. [14:57:10] cheers [14:57:45] MaxSem: Try the puppet class through labsconsole, or where you refering to something else? [14:58:06] nah, labsconsole shouldn't work:) [14:58:36] why not? [14:58:46] apmon: ... still works even after removing it from site.pp [14:59:17] hmm, ok. Perhaps I was doing something wrong then. [14:59:47] you can't apply a class that hasn't been merged in gerrit yet [14:59:51] MaxSem: Do you mind if I reboot maps-tiles1 [15:00:01] MaxSem: You can, iff you're puppetmaster__self [15:00:12] apmon, rebooting [15:00:12] puppetmaster::self* [15:00:25] mmm [15:00:26] err: Could not retrieve catalog from remote server: Connection refused - connect(2) [15:00:42] MaxSem: Is that after rebooting? [15:00:50] no, before [15:00:58] I was getting that error, but it might be because the local puppet master was killed by me [15:01:22] which is why I want to reboot the instance [15:01:42] although it might be easier to just restart puppet... ;-) [15:02:37] aha [15:02:43] now it's clearer [15:02:43] err: Could not retrieve catalog from remote server: Error 400 on SERVER: Invalid parameter day at /etc/puppet/modules/osm/manifests/tileserver.pp:42 on node maps-tiles1.pmtpa.wmflabs [15:03:12] I haven't had that issue before [15:05:49] ho do Is set the labs_mediawiki_hostname to the fully qualified hostname [15:08:38] Oren_Bochman, let me adjust your quotas so you have a public IP [15:08:48] thanks [15:11:42] would he.moodle.labs.org be an ok he.moodle.labs.org [15:12:07] labs_mediawiki_hostname ? [15:15:24] You will need an actual, existing DNS name first… it won't do much if you point labs_mediawiki_hostname to a nonexistent name. [15:15:50] And it will need to be under a domain that exists for labs already. [15:16:08] I'm confused [15:16:24] labs_mediawiki_hostname tells the wiki where it lives. [15:16:29] But it doesn't cause the wiki to live there. [15:16:34] …if that makes sense. [15:17:18] what I'm asking is - what domain would work for this [15:18:06] I guess once it works we can also set the official dns to point at the same address [15:18:55] You should start by configuring your instance with a public IP and assigning it an address, using this page: https://wikitech.wikimedia.org/wiki/Special:NovaAddress [15:19:15] That provides you with a list of high-level domains, and the ability to assign an arbitrary hostname. [15:19:29] Once you have that set, you'll enter that into labs_mediawiki_hostname when you set up the wiki [15:19:58] If you don't like the options in the list, I can create another one. It will always be under wmflabs.org. [15:20:44] (Sorry, not sure if I'm answering your question :( ) [15:21:06] MaxSem: I am now also getting the Invalid day error in puppet [15:21:25] on a new instance I just created maps-tiles4 [15:21:30] no you are very helpfull [15:24:21] Coren: I just tried it on a fresh instance and the labsconsole setting of puppet classes now indeed seems to work [15:25:05] I am reasonably sure I did the exact same thing yesterday, but oh well, as long as it works now it is fine. Thanks for your help. [15:26:01] MaxSem: After commenting out the cron script in tileserver.pp it now seems to run. [15:26:13] apmon: np. That's what they pay me the reasonably-sized bucks for. :-) [15:26:25] Be back in a few. Lunch. [15:26:29] no idea why that no longer works (or newly throws an error) though [15:27:26] hi DarTar [15:30:44] MaxSem: There are a couple of minor things I noticed about the puppet scripts. [15:30:52] yup? [15:31:43] In the mapnik datasource defininition there is a parameter , which if set forces postgresql to use a tcp connection rather than a unix domain socket [15:31:56] with that the automatic user identification no longer works [15:32:17] which means you'll need to actually set passwords and set up postgresql to use them [15:33:30] normally the easiest ist to just comment out the parameter, but as we might want to have the db on a different server than renderd, I guess that wouldn't work [15:34:54] secondly, I have removed the automatic creation of the /var/lib/mod_tile from the packages, as one might want to use a dfferent directory (or a rados store) to store tiles [15:35:45] creating the tiles directory therefore needs to be added to the puppet scripts (as other parts currently rely on its existence) [15:36:25] thirdly, it might be better to run renderd under the osm user and not under www-data. [15:36:31] in such case, you'll need to remove all the default configs that mention this dir. not worth the effort if you ask me [15:37:13] all of the configs need to be set correctly by puppet for the installation anyway [15:37:16] what's the difference between www-data nad osm? [15:37:32] nothing really, it just seems cleaner [15:37:48] but if you are OK with it, then it should work that way [15:38:49] mmm [15:38:59] other than that, I think everything worked fine [15:38:59] all the files appear to be owned by osm [15:39:21] so indeed doing that from osm might make sense [15:39:54] mod_tile, which runs in apache and therefore as user www-data needs to be able to read the files in the tile dir (assuming one is using a file based tile directory). [15:40:03] but that can be done through file permissions [15:40:12] yeah [15:40:29] but allowing just one user to write into these dirs sounds safer [15:40:46] also, there's a dependency on toolserver in load-next. I don't quite understand what it does... [15:41:11] it just translates a date to a replication sequence number [15:41:22] makes it easier to set up replication. [15:41:34] but you can get the correct sequence number through other means as well [15:42:09] e.g. looking in http://planet.openstreetmap.org/replication/minute/ to see what the correct state.txt is to start replication from [15:42:30] yeah, a dependency on third party site is dangerous [15:42:55] No issue with taking that out. [15:43:12] It is only needed once for the intitial import anyway [15:45:44] apmon: [15:45:46] "/var/lib/mod_tile/.osmosis/": [15:45:46] ensure => directory, # so make this a directory [15:45:46] recurse => true, # enable recursive directory management [15:45:58] so it should already be created [15:46:40] hmm, it failed for me yesterday. But the recurse should indeed have worked [15:47:22] Oh thanks Coren :) [15:48:05] err: /Stage[main]/Osm::Importer::Files/File[/var/lib/mod_tile/.osmosis/]/ensure: change from absent to directory failed: Cannot create /var/lib/mod_tile/.osmosis; parent directory /var/lib/mod_tile does not exist [15:48:53] somehow recurse => true doesn't seem to do what I thought recurse would do [15:49:35] lemme prepare a new commit [15:57:05] mhm, puppet docs in their shiny epicness: [15:57:21] "recurse: Whether and how deeply to do recursive management." [16:02:09] apmon, I've updated https://gerrit.wikimedia.org/r/36222 [16:07:41] looks good. [16:08:07] It might be worth making the tiles directory a variable and then use templates to substitute it [16:08:32] what for? [16:09:06] to be able to change it easily [16:09:12] e.g. when we move over to rados [16:09:37] or to move it to /data/project/tilesdir to use it as a shared directory accross instances for testing [16:10:00] but perhaps it isn't worth it [16:11:12] eh, project storage is too slow [16:12:53] as long as it works, it should be fine for testing fail-over procedures, even if it is slow [16:13:02] you just can't do load testing with it... ;-) [16:14:16] and load testing wouldn't really work in labs anyway (for one you can't get a performant full planet db on labs) [16:29:16] MaxSem: I see there is a passwords::osm::db class. Do you already use it? I guess that lives in the puppet/private directory? [16:29:31] no I don't [16:39:48] Any ideas on why a gnuplot script wouldn't run from a webtool, but would run from the command line? [17:11:44] [bz] (NEW - created by: Željko Filipin, priority: Unprioritized - normal) [Bug 47129] Upload Wizard broken at commons.wikipedia.beta.wmflabs.org - https://bugzilla.wikimedia.org/show_bug.cgi?id=47129 [17:35:50] fwilson|busy: u can haz gnuplotz! [17:38:00] yes i see :) [17:38:04] but it doesn't work from my webtool [17:38:06] strangely [17:38:32] which is http://tools.wmflabs.org/voxelbot/cgi-bin/voxel.py?dtype=graph [17:39:02] it runs a gnuplot script from recent vandalism stats, which works from the command line but not from the tool [17:39:47] fwilson|busy: Hm. Lemme check something. [17:40:46] fwilson|busy: [Thu Apr 11 16:25:28 2013] [error] [client 10.4.1.89] gnuplot: not found [17:40:46]