[00:00:02] andrewbogott: everything should be working now [00:00:08] Still - the error message is insane [00:00:15] And shouldn't break everyone's stuff [00:00:34] And if you "rev_user = (SELECT user_id FROM user WHERE user_name = ?)"? [00:00:36] Damianz: Maybe the request /does/ time out and the php script gets wedged in a crazy mode? [00:01:06] andrewbogott, Damianz: quick code review? https://gerrit.wikimedia.org/r/#/c/64868/ [00:01:07] scfc_de: comparing user isn't all that much faster from user_text [00:01:15] I'd expect it to die, but it doesn't seem to [00:01:37] Damianz: Lemme try to figure out why the query is slow. It probably shouldn't be. [00:01:49] Ryan_Lane: Looks ok to me [00:01:56] +1? :) [00:01:58] Coren: I was trying to do an explain, but I can't lock the table [00:02:01] I'll +2/merge [00:02:14] (I've already tested it [00:02:15] ) [00:02:21] wait [00:02:22] shit [00:02:25] wrong thing? [00:02:35] https://gerrit.wikimedia.org/r/#/c/64868/ [00:02:44] no. right thing [00:02:53] I have too many tabs open [00:03:05] Damianz: That's normal, and pretty much unavoidable. [00:03:19] 2110 < just a little time [00:04:10] Ryan_Lane: But do you have more than me? :p [00:04:23] a930913: more what? [00:04:29] Ryan_Lane: Tabs. [00:04:33] oh [00:04:33] heh [00:04:36] Damians: select count(distinct rev_page) from revision_userindex where `rev_user_text` = 'Cydebot'; -- takes 11s [00:04:40] I don't know. I only had 40 open [00:05:02] Is 1.7 million pages likely? [00:05:13] Ryan_Lane: Only 40? I have that in just one window. :p [00:05:17] :D [00:05:26] Hm. seems about right. [00:05:27] Coren: Possibly [00:05:42] 3646663 without distinct [00:05:50] and it did seem fast, but there's doezens in show processlist with hundreds of seconds against them [00:06:11] Damianz: Note the slightly distinct query I use. [00:06:28] you're avoiding a sub select [00:06:30] On 'coren' it only takes 0.04 seconds. [00:07:24] andrewbogott: have you tested this yet? https://gerrit.wikimedia.org/r/#/c/64303/ [00:07:29] 1sec, we'll see [00:07:52] Ryan_Lane, I can log in to nova-precise2 but it's still messed up in various ways. But… I need to go soon so don't fix it now on my account :) [00:08:03] heh [00:08:10] Damianz: I think your subquery created a temporary table. :-) [00:08:17] I've tested that sudoers patch enough to make sure that the right stuff is getting into ldap. I haven't tested the function of the actual policies. [00:08:19] Would make sense [00:08:45] well, if you tested the correct stuff is going in, that's fine [00:09:00] hm. I wonder why this is timing out [00:09:09] Damianz: Did you change your query? because it's behaving MUCH better now. [00:09:14] yes [00:09:28] Just going to kill these queries, but it seems ok - hitting it wiht 1k of requests [00:09:46] Damianz: Looks like some evil interaction with php and timeout of the query. [00:10:34] Warning: There is 1 user waiting for shell: Shawnanthony420 (waiting 10 minutes) [00:10:52] andrewbogott: fixed [00:11:00] it was glance [00:11:07] Coren: Yup - it's maxing out at like 4/5 sec now... much better [00:12:46] * Damianz wonders how many more ways he can break Coren's servers in an evening [00:13:04] Damianz: It's good that you are stress testing it. :-) [00:13:25] Damianz: /me is still trying to figure out /why/ apache started going bonkers withe 75% free ram though. [00:13:30] Damianz: Just in time to catch the second vandalism peak :) http://bit.ly/180Xhfo [00:14:52] a930913: =D [00:15:54] Without being scientific and doing the whole graphing thing, I'm pretty sure this db is faster/more up to date than ts... seems more edits are going past the point of bailing out because it can't get the data. [00:16:19] Does that graph suggest that a large percentage of vandalism is done by Americans at work? [00:17:08] hm. I guess I shouldn't deploy this OpenStackManager change right before I leave, eh? [00:17:09] Then dips when workers go home, until they arrive home and vandalise a bit more. [00:17:16] especially seeing as that I'm flying tomorrow.... [00:17:41] Ryan_Lane, you mean the sec group thing, or the sudoer thing, or both? [00:17:41] Damianz: It'd make sense, it's much closer to the masters. [00:17:47] andrewbogott: both :) [00:17:57] I guess neither one will really break things [00:17:59] I can deploy tomorrow when I have a day ahead of me [00:18:10] oh. well, I'll deploy it now, then :) [00:18:13] if you'll be around to revert [00:18:18] Tomorrow I will [00:18:21] cool [00:18:30] Ryan_Lane: You're like american.. you get wireless on planes [00:18:44] Damianz: not to AMS I don't ;) [00:18:46] Damianz: Aha. A quick google sez: You're breaking the PHP memory limit. [00:18:57] Ryan_Lane, btw, the instance status pages should be working again, let me know if you see anything amiss there [00:19:02] Coren: The php memory limit isn't per /process/? [00:19:08] andrewbogott: sweet. thanks [00:19:25] Oh yeah it's AMS this weekend isn't it [00:19:33] Hm, no real instances on nova-precise2 yet :( [00:19:47] Damianz: It is, but maybe something odd goes on in the process when the script wedges on a timed out connection? [00:19:54] andrewbogott: it still errors? [00:20:16] yeah, although it was marked as 'building' briefly. I haven't investigated [00:20:39] * Ryan_Lane nods [00:20:41] Coren: It's like what you'd see with mod_php where it fills up the worker pool and maxes out the apache slap... but it's suphp, so should be pretty much true processes... and I've only ever seen that with ulimit style restrictions before [00:20:46] well, at least memory won't be it now :) [00:20:51] should make it easier to debug [00:21:13] Oh god gerrit, my poor inbox [00:21:29] 15 emails is just not needed [00:21:31] Damianz: I checked; that's not what you're hitting. I suspect a bug with the query timing out. [00:22:01] Mental note: Write custom c++ server to accept xml, return xml and query mysql [00:22:04] * Damianz rofls [00:22:22] It's weird.. but yeah [00:23:01] The stranger question is, why did it break all the users... not jsut the one since they should be forked then setuid... something must be really screwed [00:23:59] Warning: There is 1 user waiting for shell: Shawnanthony420 (waiting 24 minutes) [00:32:20] Damianz: I'll just keep an eye on it. Besides, activating the second backend is overdue. [00:33:29] going to split the apps betwean then with cold failover if required? [00:35:00] Damianz: That's the objective. [00:37:28] Warning: There is 1 user waiting for shell: Shawnanthony420 (waiting 37 minutes) [00:41:34] Coren: PHP Fatal error: Out of memory (allocated 262144) (tried to allocate 523800 bytes) in Unknown on line 0 [00:41:41] Only a few times, but ties up with something weird happening memory wise [00:41:53] * Damianz dislikes php_errors.log [00:42:01] Damianz: Why? [00:42:15] What's wrong with error_log with everything in it :P [00:42:48] Damianz: It discloses too many things atm. I'm working on a new way of doing things. [00:44:10] The problem with using php_errors (assuming it's the error log set as a php option - looks like it) is php has to actually run for it to error :( [00:45:04] It is, and you're correct. Which is why I will look for an alternative. [00:45:52] I may have to please Ryan and go virtual-host-per tool; which has other inconvenients. [00:46:20] dunno, could do .toollabs.org or someting... no ll though, that looks stupid [00:46:56] Oh, no need for dns trickery. [00:47:09] Still, that's for the After-Amsterdam [00:47:45] Hmm... I guess you could do multi host mod_userdir style, never been that mental [00:50:43] Damianz: I just improved error reporting a lot about php not starting. :-) [00:51:11] :D [00:58:34] Warning: There is 1 user waiting for shell: Ilfeld (waiting 0 minutes) [01:00:00] Coren, petan: Could you please 'for JOBID in 81662 81672 81648 81659 81660 81649 81658 81661 81669 81670; do qacct -j "$JOBID" | mail -s "$JOBID" scfc; done' on tools-master? TIA. [01:00:27] scfc_de: You keep forgetting the ';done' :-) [01:01:19] Coren: Hmmm? Do you mind the space in "; done"? :-) [01:01:56] Oh, no. Ignore me. [01:01:58] (I don't care if its one mail per job or everything.) [01:02:08] I just realized that I did the other thing last time. :-) [01:02:28] scfc_de: {{DONE}} [01:02:37] Coren: Thanks :-). [01:03:07] What pager does mail use on Ubuntu? It sucks for people used to less. [01:03:37] scfc_de: I think it's more. If I were you, I'd use alpine [01:03:45] and not mail. :-) [01:04:04] you can usually export PAGER=/usr/bin/less though [01:04:47] Well, / is now on -02 [01:04:59] Seems to work. [01:05:17] Damianz: Gonna move clubot there [01:05:32] If I got a labs account, I'd get a shell to run my bots in, right? [01:05:52] Coren: Yep, works nicely. alpine -- heaven no. Starts with "Creating subdirectory "/home/scfc/mail" where Alpine will store its mail folders." -- I don't want that, just read mail. [01:06:30] a930913: You need to be a member of the Tools project as well. [01:06:49] (Don't ever run anything on Labs bastion :-).) [01:07:05] Bleh, is it worth all the effort? [01:07:51] What effort? What's your Labs username? [01:07:52] Damianz: Tell me if that broke cluebot, but I don't expect it should. [01:08:17] Coren: Looks good here. [01:08:43] scfc_de: I don't have one. [01:09:02] I'm currently running my stuff on my computer. [01:11:02] a930913: Just follow the instructions on https://wikitech.wikimedia.org/wiki/Nova_Resource:Tools/Help#Getting_Access. You'll get access to replicated databases, a 24/7 platform, and lots of other developers you can collaborate with -- so definitely worth the effort :-). [01:11:59] Warning: There is 1 user waiting for shell: Ilfeld (waiting 13 minutes) [01:14:41] scfc_de: But they want RL info, no? :( [01:15:27] I don't, never concerned me. [01:15:31] *know [01:15:49] scfc_de: Don't you have to give your real name oe? [01:16:32] I don't see a rule that requires this. [01:16:38] http://www.mediawiki.org/wiki/Wikimedia_Labs/Terms_of_use [01:19:06] a930913: First I hear of this. [01:23:37] Clearly I am a bot, because these are some of the worst capchas I've seen >.> [01:25:31] Warning: There is 1 user waiting for shell: A930913 (waiting 0 minutes) [01:30:52] a930913: I've preemptively added you to the tools project as well. [01:35:34] Coren: \o/ [01:35:38] (Doing TAR now.) [01:37:02] As a reminder, in case you haven't read all the doc, you don't want to run tools from your personal account, but create service groups for them. :-) [01:37:49] I think I might have to read the docs :p [01:37:53] Can I mosh in? [01:38:19] Yep [01:38:56] "No Nova credentials found for your account."? [01:39:00] Warning: There is 1 user waiting for access to tools project: A930913 (waiting 1 minutes) [01:39:30] I'm not sure preemption worked :p [01:41:22] Hmm, I don't need to mosh in, I used the key on my server. :) [01:48:07] Coren: Do you need to gerfingerpoken something more so I can make a tool? [01:52:24] Warning: There is 1 user waiting for access to tools project: A930913 (waiting 15 minutes) [01:53:06] a930913: Do you get an error message? [01:54:36] a930913: You may have to log off and back on again. [01:54:48] a930913: In fact, I'm pretty sure you have to. :-) [01:59:17] \o/ [02:00:15] So I need to make the account "local-bracketbot"? [02:01:07] No, "Add service group" ^, right? [02:05:45] Warning: There is 1 user waiting for access to tools project: A930913 (waiting 28 minutes) [02:08:40] a930913: No, the name should be just "bracketbot". [02:09:56] scfc_de: I think it stripped it for me? [02:12:12] So I should just upload my script into the labs and run "jstart script"? [02:14:04] a930913: In the most basic form, that should be enough. [02:14:33] (If it's continuously running and should be restarted if it stops.) [02:15:20] Do I get a high port to open publicly? [02:16:11] Because my bot does its reporting via an irccat. [02:19:06] Warning: There is 1 user waiting for access to tools project: A930913 (waiting 42 minutes) [02:21:12] There bots that report on IRC, you should probably copy their behaviour. Someone still around? Damianz? petan? addshore? [02:22:56] scfc_de: My cat listens on my computer for UDP12345 messages and prints them to IRC. [02:23:26] So if I moved the cat, I'd want to be able to send from my computer too. [02:25:27] Don't you have an IRC client?! [02:26:42] scfc_de: Eh? [02:27:14] scfc_de: As in I can cat reports to the cat. [02:27:40] And it will cat it to IRC. [02:28:46] a930913: Ah, so you want to run the bot on your computer in parallel? Then why don't you add IRC support on it there as well? Sending data from the bot by another protocol to another daemon seems like another potential point of failure to me. [02:30:34] Well, different bot stuff. The lovely thing about UDP is that it's best effort, and failure doesn't break things. [02:31:50] Though actually... If the daemon is run on the grid, how do you find it? [02:32:27] Warning: There is 1 user waiting for access to tools project: A930913 (waiting 55 minutes) [02:33:45] I don't know, that's why I would keep it all in one place. But I don't have experience with bots that report to IRC. [02:38:38] a930913: You don't really; inbound to bots aren't supported (though outbound are) [02:39:05] a930913: So a bot can connect to IRC for instance. [02:39:56] An ugly hack I can think of is to make an HTTP carrier proxy it :p [02:40:36] Coren: I want a bot to connect to IRC on the labs, and listen on a port for incoming messages. [02:41:26] The latter isn't supported; that's too much of a security liability. OTOH, you can place a small script in your public_html that'd allow remote control via an RPC call. [02:41:46] Much more robust scheme, too. [02:45:47] Warning: There is 1 user waiting for access to tools project: A930913 (waiting 68 minutes) [02:46:17] Coren: But how can I find my IRC bot from the public_html script? [02:47:26] a930913: The simplest scheme is to communicate through the local filesystem. You can pass information through files, the home is shared between all the nodes. [02:47:44] Coren: Filesystems, really? [02:48:04] a930913: Hey, it's well proven. Just plop your ip and port in a file for the script to find. [02:48:58] Meh, fine. [02:49:38] If I ":~$ become bracketbot" should it complain about sudo? [02:49:45] Coren: Could you explain the difference between logging and logging_userindex again? [02:49:55] a930913: What's the error message? [02:50:09] scfc_de: "sudo: sorry, a password is required to run sudo" [02:50:12] a930913: Did you create your tool with local- in the name? [02:50:25] Coren: Might have :/ [02:50:51] Known bug. :-( It will only have partially stipped it and the tool account is broken. I'll delete it and recreate it for yu. [02:51:40] * a930913 hands Coren an extra large double clawed hammer to fix it with. [02:52:05] a930913: You'll have to log off and back on (from -login) for the new group to take effect. [02:52:50] Still complaining. [02:53:11] Oh, wait. I might have forgotten a step. [02:53:48] Do I install stuff like pywikibot and twisted locally? [02:58:59] a930913: Try this. [02:59:08] Warning: There is 1 user waiting for access to tools project: A930913 (waiting 82 minutes) [02:59:27] wm-bot: You are insane. [02:59:27] Hi Coren, there is some error, I am a stupid bot and I am not intelligent enough to hold a conversation with you :-) [02:59:32] Coren: Nope. [02:59:43] You did log off? I mean since last time? :-) [03:00:05] Logged off after you said to try it. [03:00:25] * Coren is a moron. [03:00:33] Perhaps, just perhaps, I should put you in the group? :-) [03:01:05] Hmm, maybe.. :-) [03:02:36] Ah, and it's cached locally by the ldap daemon. How... uncouth. [03:06:21] Something odd is going on. [03:07:26] Coren: BTW, is the local- issue tracked somewhere? [03:07:55] Hm. I honestly don't know. I know andrew is aware of it, but I haven't noticed if there's a bz for it [03:08:38] I'll file one just in case so it doesn't get lost. [03:11:53] a930913: Something is /really/ odd about your account in a way I haven't yet figured out. For some reason, it seems unable to actually give you membership in any tool automatically. [03:12:32] Warning: There is 1 user waiting for access to tools project: A930913 (waiting 95 minutes) [03:12:59] Or nslcd is caching the hell out of something. [03:13:23] Coren: I do apologise for the inconvenience. [03:13:41] * a930913 mutters something about knowing it was a bad idea. [03:13:49] a930913: No, no, thanks actually. The faster those odd problems crop up, the faster they get squished. :-) [03:14:11] * a930913 hands Coren a bigger hammer. [03:24:40] a930913: I'd like to try something else. Do you have an alternate name you'd find acceptable for your tool? [03:25:09] For some reason, it's not able to fully build it with that name (and I can't figure out what could possibly be wrong with it) [03:25:21] Looks like it got partly created and broke. [03:25:50] Go create a new group, and I'll move and chown the files. [03:25:57] Warning: There is 1 user waiting for access to tools project: A930913 (waiting 108 minutes) [03:26:09] Coren: Should I make a generic name for running all my private bots? [03:26:14] If that works, I'll have Andrew (the maintainer of that code) look into why it broke so it doesn't break again. [03:26:51] a930913: Well, depends. Some people prefer that, some prefer separating them so they can hand off maintenance or add helpers differently depending on which tool. Your pick. [03:28:37] D: [03:28:51] Doesn't let me choose 930913. [03:29:00] No, has to start with a letter. [03:29:23] * a930913 rages against that being so common. [03:29:40] It avoids ambiguity with user IDs. [03:30:06] Coren: To confirm, the 500 thrown is the confirmation page, right? [03:30:19] ... that shouldn't happen actually. [03:30:32] Lol. [03:30:46] ... and the group you just created is broken in the same way(!) [03:31:00] What in /blazes/ is wrong with your account? [03:31:16] * a930913 is just special. [03:31:22] * a930913 headdesks. [03:32:31] * Coren sees nothing wrong with it. [03:34:50] Hm. Nope, it's not your account. Service groups I create are just as broken. [03:35:02] Wikitech is teh br0ken tonight. [03:35:21] * Coren pokes the right people [03:35:28] Sorry for the inconvenience. [03:38:30] * Coren tries to see if he can find something obvious. [03:38:39] Not my code, though. It'd have to be hella obvious. [03:39:18] Warning: There is 1 user waiting for access to tools project: A930913 (waiting 122 minutes) [03:41:31] Hm. I think I might actually have found it. [03:41:49] :o [03:42:15] * a930913 hands Coren a machete to cut the blighter out with. [03:44:33] Yep. It's a bugfix that introduced a new bug. How fun. [03:46:54] * a930913 hands Coren a printed version of wikipedia to squash the bug with. [03:47:36] Coren, still up? [03:47:51] andrewbogott: I am. Check out https://gerrit.wikimedia.org/r/#/c/64880/1 [03:49:04] andrewbogott: Tried it by hand directly on Wikitech. Fixes it. [03:49:19] Makes sense. [03:49:43] * Coren is just glad the error_log pointed at the right direction. [03:49:44] Might be good to make it case-insensitive. Also, should probably add actual error handling to the bit that was crashing when the lookup failed... [03:49:53] Are you in deep enough to do that or do you want me to take over the patch? [03:50:15] And, does that fix both problems you mentioned in the email? (service groups + sudoers) [03:50:36] Yes, the service group failed to create properly because it broke when trying to add the sudoers. [03:50:57] makes sense. [03:51:07] Feel free to take it over, that was just a panicky hotfix. :-) [03:51:50] ok [03:52:07] Do you need me to stick around? I've been up for ~20h now. :-) [03:52:51] Warning: There is 1 user waiting for access to tools project: A930913 (waiting 135 minutes) [03:52:55] FIY, the break was at getFullyQualifiedDisplayName() that broke because the $hostobj was null. [03:53:19] Coren: So what was special about me that broke it? [03:53:38] Just guarding against /that/ would work too, I suppose, though you'd still want to guard against trying to be smart with 'all' [03:53:51] a930913: Nothing, you're just the first to stumble upon it. :-) [03:53:54] Yep, should do both [03:54:23] Coren: What changed? I was the first since that bugfix? [03:54:46] a930913: Or the first to notice. Dunno when that code was last updated -- I expect Andrew could tell you. [03:55:44] andrewbogott: I can haz sleepz? :-) [03:55:54] Yeah, I'll fix and merge. [03:55:58] Coren: Good night! :-) [03:56:03] Good tight, Andrew, and thanks. [03:56:15] a930913: You should be all set once that patch goes in. Good luck. [03:56:20] a930913, it was caused by a patch that Ryan merged right before I went out for dinner, a couple of hours ago. [03:56:35] (Um, a patch I wrote, I should add :) ) [04:06:12] Warning: There is 1 user waiting for access to tools project: A930913 (waiting 149 minutes) [04:07:55] andrewbogott: Is it patched now? [04:08:01] not yet [04:13:25] a930913, want to try it now? [04:13:47] * a930913 cues the explosions. [04:14:20] Well, it threw the 500 again. [04:14:25] hm, ok. [04:14:44] * andrewbogott returns to the test box [04:15:11] Nope, still the sudo error. [04:15:55] How about now? [04:16:56] andrewbogott: Making another service group, or becomeing? [04:17:10] Um… making a service group. [04:17:16] If that fails then becoming will definitely not work [04:18:02] 500 [04:19:07] curious [04:19:44] Warning: There is 1 user waiting for access to tools project: A930913 (waiting 162 minutes) [04:28:01] a930913: OK, once more? [04:30:56] 200 [04:31:52] that's good, right? [04:31:55] andrewbogott: Works :) [04:32:05] ok, lemme tie this up... [04:32:05] andrewbogott: HTTP 200 is OK. [04:32:20] Yeah, but just 'cause it returns 200 doesn't mean it did anything :) [04:32:39] andrewbogott: Could you remove the tests and bracketbot, so I can make a proper bracketbot. [04:32:53] sure, just a second [04:33:13] Warning: There is 1 user waiting for access to tools project: A930913 (waiting 176 minutes) [04:33:24] andrewbogott: Well, it was a note of improvement on the 500 :p [04:36:39] a930913: OK, cleaned up. [04:40:35] andrewbogott: \o/ [04:46:46] Warning: There is 1 user waiting for access to tools project: A930913 (waiting 189 minutes) [05:00:09] Warning: There is 1 user waiting for access to tools project: A930913 (waiting 203 minutes) [05:13:30] Warning: There is 1 user waiting for access to tools project: A930913 (waiting 216 minutes) [05:26:50] Warning: There is 1 user waiting for access to tools project: A930913 (waiting 229 minutes) [05:40:11] Warning: There is 1 user waiting for access to tools project: A930913 (waiting 243 minutes) [05:53:35] Warning: There is 1 user waiting for access to tools project: A930913 (waiting 256 minutes) [06:07:00] Warning: There is 1 user waiting for shell: Fantasticfears (waiting 4 minutes) [06:07:01] Warning: There is 1 user waiting for access to tools project: A930913 (waiting 269 minutes) [06:20:20] Warning: There is 1 user waiting for access to tools project: A930913 (waiting 283 minutes) [06:33:53] Warning: There is 1 user waiting for access to tools project: A930913 (waiting 296 minutes) [06:47:18] Warning: There is 1 user waiting for access to tools project: A930913 (waiting 310 minutes) [07:00:47] Warning: There is 1 user waiting for access to tools project: A930913 (waiting 323 minutes) [07:14:07] Warning: There is 1 user waiting for access to tools project: A930913 (waiting 337 minutes) [07:27:32] Warning: There is 1 user waiting for shell: Yonjah (waiting 8 minutes) [07:27:33] Warning: There is 1 user waiting for access to tools project: A930913 (waiting 350 minutes) [07:36:37] petan: bots-login doesn't allow outbound network connections? [07:37:06] liangent@bots-login:~$ sftp nightshade.toolserver.org [07:37:06] ssh: connect to host nightshade.toolserver.org port 22: Network is unreachable [07:37:07] Couldn't read packet: Connection reset by peer [07:40:53] Warning: There is 1 user waiting for shell: Yonjah (waiting 21 minutes) [07:40:54] Warning: There is 1 user waiting for access to tools project: A930913 (waiting 363 minutes) [07:54:18] Warning: There is 1 user waiting for shell: Yonjah (waiting 35 minutes) [07:54:19] Warning: There is 1 user waiting for access to tools project: A930913 (waiting 377 minutes) [08:07:51] Warning: There is 1 user waiting for shell: Yonjah (waiting 48 minutes) [08:07:52] Warning: There is 1 user waiting for access to tools project: A930913 (waiting 390 minutes) [08:21:20] Warning: There are 2 users waiting for shell, displaying last 2: Yonjah (waiting 62 minutes) Damzow (waiting 7 minutes) [08:21:21] Warning: There is 1 user waiting for access to tools project: A930913 (waiting 404 minutes) [08:23:30] liangent it should allow it [08:23:50] liangent I often connect to foreign servers, I would rather say toolserver is down [08:25:25] !tr A930913 [08:25:25] request page: https://wikitech.wikimedia.org/wiki/Nova_Resource:Tools/Access_Request/A930913?action=edit talk page: https://wikitech.wikimedia.org/wiki/User_talk:A930913?action=edit§ion=new&preload=Template:ToolsGranted link: https://wikitech.wikimedia.org/w/index.php?title=Special:NovaProject&action=addmember&projectname=tools [08:25:31] !rt A930913 [08:25:31] http://rt.wikimedia.org/Ticket/Display.html?id=A930913 [08:25:42] lol [08:27:02] Coren|Sleep if you grant someone access to tools project, you need to also flag the request as completed, or the bot will keep complaining... :/ [08:27:25] if you don't like this design you can of course invent a better one [08:28:19] a page with three iframes, corresponding to each of those to-be-edited pages? :p [08:32:01] !petan- [08:32:01] There are multiple keys, refine your input: petan-build, petan-forgot, [08:32:05] !petan-build [08:32:06] make -j `getconf _NPROCESSORS_ONLN` deb-pkg LOCALVERSION=-custom [08:33:37] petan: I can ssh to nightshade at ts from my own computer [08:34:52] Warning: There are 2 users waiting for shell, displaying last 2: Yonjah (waiting 75 minutes) Damzow (waiting 20 minutes) [08:37:51] liangent let me try [08:39:19] yes it is weird but connection to other hosts work [08:39:24] just toolserver is broken :/ [08:39:34] maybe ask Ryan or someone who is responsible for network [08:48:21] Warning: There are 2 users waiting for shell, displaying last 2: Yonjah (waiting 89 minutes) Damzow (waiting 34 minutes) [09:01:46] Warning: There are 2 users waiting for shell, displaying last 2: Yonjah (waiting 102 minutes) Damzow (waiting 47 minutes) [09:10:05] !rq Yonjah [09:10:05] https://wikitech.wikimedia.org/wiki/Shell_Request/Yonjah?action=edit https://wikitech.wikimedia.org/wiki/User_talk:Yonjah?action=edit§ion=new&preload=Template:ShellGranted https://wikitech.wikimedia.org/wiki/Special:UserRights/Yonjah [09:10:10] !rq Damzow [09:10:10] https://wikitech.wikimedia.org/wiki/Shell_Request/Damzow?action=edit https://wikitech.wikimedia.org/wiki/User_talk:Damzow?action=edit§ion=new&preload=Template:ShellGranted https://wikitech.wikimedia.org/wiki/Special:UserRights/Damzow [09:15:11] Warning: There is 1 user waiting for access to tools project: Tb (waiting 5 minutes) [09:16:22] !tr Tb [09:16:22] request page: https://wikitech.wikimedia.org/wiki/Nova_Resource:Tools/Access_Request/Tb?action=edit talk page: https://wikitech.wikimedia.org/wiki/User_talk:Tb?action=edit§ion=new&preload=Template:ToolsGranted link: https://wikitech.wikimedia.org/w/index.php?title=Special:NovaProject&action=addmember&projectname=tools [09:18:01] Coren|Sleep: how the replica users are being created? [09:18:05] is it automatic? [09:33:15] Do I install stuff like pywikibot and twisted locally? [09:33:44] define locally [09:34:38] In my/the tools space. [09:34:47] you can install it locally to /tmp or you can install it to your home which is shared on all instances, or you can install it to /shared which is shared too. everywhere. [09:35:34] locally == local to instance || locally == local to user? [09:37:28] Okay, so pywikibot is in /shared. [09:38:49] And twisted is installed. [09:38:55] That's me for now. [09:38:59] :o [09:39:19] But let's say that I wanted "pip install txws" [09:39:27] What's the right way to do it? [09:40:03] request it to be converted to .deb package OR convert it yourself to .deb package and request it to be installed project wide [09:40:45] !pypi [09:40:45] https://pypi.python.org/pypi/stdeb#the-commands [09:41:26] Oh cool. [09:42:08] petan: Any restrictions on the internal ports I use? [09:42:18] depends [09:42:25] (High) [09:42:32] instances aren't accessible from outside, except for -dev -login and -proxy [09:43:01] otherwise there are no restrictions I know of as long as you need to access these ports on local instances only [09:50:05] petan: Just so something doesn't blow up, talk me through my first job? (A bot for logging to IRC.) [09:51:12] I just run "jsub -once python bot.py"? [09:52:57] a930913: if it needs to run continuously, use -continuous [09:53:10] a930913: in any case, add -N jobname; otherwise it will use 'python' as job name [09:53:33] e.g. jsub -once -N myawesomeircbot python bot.py [09:53:50] Is the jobname unique to me? [09:56:02] yes [09:57:15] a930913 I recommend you to cron a job that would check if your bot is running and if not it start it [09:57:25] because IMHO the grid engine is very much not reliable in this [09:58:20]