[00:02:10] YuviPanda: any idea why I can't create an nfs-using instance in proxy-project? Is it somehow more complicated than checking the box and rebooting? [00:02:21] I've done it three times now, and each time I get no mounts. [00:02:36] andrewbogott: no idea, but IIRC NFS itself is a bit borked? [00:02:53] Oh, I thought it was only briefly broken. [00:03:02] I just want to create a new instance to test the new .debs and puppet but... [00:03:04] can't! [00:03:20] andrewbogott: hehe :D [00:03:22] andrewbogott: poke Coren? [00:03:26] andrewbogott: also do you *need* NFS? [00:03:44] Well, the current labsdebrepo you've been using is on NFS [00:04:04] ah [00:04:04] right [00:04:07] I guess I can ssh it into an equivalent repo [00:04:14] andrewbogott: there's already a test instance you can use? [00:04:26] andrewbogott: oh, nevermind, that is its own puppetmaster [00:04:26] ? [00:07:27] * Coren pokes back! Poink! [00:08:03] andrewbogott: poke Coren about nfs being broken! [00:08:54] Coren, I can't get any new NFS-mounting instances to behave. For example proxy-abogott-4 [00:08:59] andrewbogott: Try again? [00:09:05] (You may need to kick autofs) [00:09:14] wouldn't a reboot do that? [00:09:40] Yeah, but you probably don't have to. Oh, I mean kick autofs again, I just tried something. [00:10:01] What's the name of the project? [00:10:17] Yep, seems better now. [00:10:20] proxy-project [00:11:06] Yeah, I see what the problem is; I'll look into it tomorrow morning. The manage-nfs-volumes-daemon doesn't seem to be actually able to do the exportfs once it's done writing the exports file. [00:11:23] Just doing a forcible exportfs -ra fixed it. [00:11:41] ok, so, I should expect the same failure on future instances? [00:11:42] For now? [00:11:55] I think only future new volumes. [00:12:23] ok, that probably won't hit me then. thanks! [00:13:04] are we ready to start using NFS for everything? [00:13:10] has it been stable enough for that? [00:13:22] Ryan_Lane: I would have said yes were it not for that newfound buglet. [00:13:31] which new bug? [00:13:56] The manage volume daemon doesn't seem to completely export new volumes it creates. Easily worked around, but I'd rather fix it first. [00:15:17] Otherwise, it's been running for a while without so much as a hiccup; and performance is, well: http://ganglia.wikimedia.org/latest/graph.php?r=week&z=xlarge&h=labstore4.pmtpa.wmnet&m=cpu_report&s=descending&mc=2&g=cpu_report&c=Labs+NFS+cluster+pmtpa [00:15:50] "Woosh" seems to be the correct word. [00:22:53] * Ryan_Lane nods [00:32:47] (03PS1) 10Andrew Bogott: s/dynamic-proxy-api/dynamicproxy-api [labs/invisible-unicorn] - 10https://gerrit.wikimedia.org/r/91541 [02:26:06] Coren, new instance, same problem -- proxy-abogott-6, project 'project-proxy' [02:26:36] andrewbogott: Try again? [02:26:50] If that works, it's definitely the exportfs that fails when done from the daemon. [02:26:56] yep, worked [16:15:39] YuviPanda: yet another packaging favor? Can you see if you can make the unicorn work with python-flask-sqlalchemy_0.9.1-1_all.deb that's in /data/project/repo? [16:15:48] Using that vs. 1.0 makes my life quite a bit easier [16:16:13] andrewbogott: moment, eating. i'll check in a bit [16:16:18] thanks [16:18:10] Where are editor rights held in the repdb? [16:27:35] (03PS1) 10Andrew Bogott: Further attempts to make this work with setuptools [labs/invisible-unicorn] - 10https://gerrit.wikimedia.org/r/91620 [16:30:18] andrewbogott: nope, doesn't work [16:30:32] And can't be worked around? [16:30:48] andrewbogott: https://dpaste.de/m8Yw [16:31:09] Yeah, that's what I saw, which is why I asked if you could make it work :) [16:31:22] andrewbogott: I think it's a problem because of an older *sqlalchemy* itself [16:31:25] so that'll need packaging [16:32:06] andrewbogott: no, I think not. 0.8 made a good number of changes [16:34:06] 0.8? [16:34:14] andrewbogott: of sqlalchemy [16:34:25] current one is 0.7.x, and I think that's the source of the error [16:34:33] the declaritive stuff was completely overhauled [16:35:24] So an even older version of flask-sqlalchemy might help... [16:35:55] andrewbogott: not really. Flask-SQLalchemy is just glue code between flask and... sqlalchemy. the error if you notice is coming from SQLAlchemy itself [16:36:04] the flask layer isn't the problem [16:36:32] andrewbogott: I guess when I did pip install Flask-SQLAlchemy it got with it a newer version of sqlalchemy, hence causing things to actually work [16:36:35] I don't understand… what is the problem then? If you think all the lower components are compatible... [16:36:56] andrewbogott: the code isn't? [16:37:11] andrewbogott: the code is not compatible with sqlalchemy < 0.8, apparently [16:37:17] Yeah, but we have the power to /change/ the code. I thought you said before that that couldn't be done [16:37:27] oh sure, it's just a fair bit of work [16:37:36] i'll have to redo the db layer almost completely [16:38:08] for some definiteion of fair bit of work - couple of hours or so after hunting through docs, perhaps. [16:38:10] oh, hm. [16:38:43] Well, I will dive back into dependency hell for the moment, see if I can get that to work with < 100 packages [16:39:13] heh, okay! [16:50:05] hello [16:51:01] i have a problem when i request a toolserver account [16:52:52] Ahmad_Sammour: what's that? [16:53:08] Coren: can you do that… thing you did again? With nfs? [16:54:43] when i go to this link : http://goo.gl/a13QYG , the page said that i cant edit the page. [16:56:20] are you logged in? [16:56:30] no [16:57:39] Hm… I think you need a labs account, and to be logged in first. Try this: https://wikitech.wikimedia.org/w/index.php?title=Special:UserLogin&type=signup&returnto=Main+Page [16:58:35] whats that mean:"Instance shell account name" [16:59:09] Ahmad_Sammour: You'll have a username on the wiki, and also a username on the actual servers you run your tool on. [16:59:19] 'shell account name' is your login name on the servers. [16:59:23] In… the shell. [17:00:41] an error messege: "There was either an authentication database error or you are not allowed to update your external account. " [17:01:05] That means that either you already have an account with that name, or you used invalid characters in your shell or wiki name. [17:01:07] Probably :) [17:01:36] andrewbogott: That shouldn't be necessary anymore; but your daemon runs on a 180s cycle. [17:02:19] What is the name of the project and the IP of the instance that's troublesome? I want to see if it should have been exported. [17:02:30] Coren: on instance proxy-abogott-7 I can't log in as myself, and when I try to acces /data/project it hangs, possibly forever [17:03:08] What project is this? [17:03:17] project-proxy [17:04:20] That's odd; I do see it exported properly. [17:04:36] So maybe this time the problem is local to the instance somehow [17:04:44] Lemme go look. [17:04:51] Ahmad_Sammour: any luck? If you think you might already have an account you can try a password reset [17:05:08] no i dont have [17:05:28] i tried probably 20 names!!! [17:06:13] Ahmad_Sammour: what username and shell name are you using? [17:06:30] for example? [17:06:35] andrewbogott: root 7784 0.0 1.0 266388 22432 ? Ssl 16:13 0:00 /usr/sbin/glusterfs --volfile-id=/project-proxy-home [17:06:35] This thing is running gluster. [17:06:59] I just switched it over and restart autofs. [17:07:03] Are there other steps somehow? [17:07:17] I'll reboot it if you're clear [17:07:22] If you didn't reboot, did you run puppetd first? :-) [17:07:35] ran puppet several times [17:07:45] I put "ASammour" in username and "AMSammour" in Instance shell account name. [17:08:00] Huh. Autofs is annoyingly finicky; I'm guessing it'd need a reboot if it ever touched gluster. [17:08:05] Feel free to reboot it [17:08:06] Try just lowercase for your shell name [17:08:07] ok [17:09:26] *sigh* Coren, seems to be working. I guess I should just be rebooting more often :) [17:09:29] andrewbogott: The lesson here is that autofs doesn't survive its configuration changing radically without leaving crap behind and a reboot is necessary. [17:09:37] yep [17:10:03] it's work! [17:10:05] It's an annoyance that'll become moot when we switch to NFS by default. [17:10:30] Ahmad_Sammour: ok, now you should be able to log in and fill in the form that you started out wanting to fill in :) [17:11:15] Ahmad_Sammour: meanwhile I will turn on your 'shell' right so that you can actually log into systems when the time comes. You'll get a confirmation email about that. [17:13:38] Coren: random apt question… I have a newer version of a package in a local repo, but apt is insisting on grabbing the upstream version and saying that it's the latest. Any idea? Does this have to do with some kind of priority marking in sources.list maybe? [17:15:26] Coren: how long does it take approximately until new project requests are accepted? (https://wikitech.wikimedia.org/wiki/New_Project_Request/Merl-tests) [17:15:28] I want the account to use it in my bot, should i put this reason or what? [17:15:47] andrewbogott: There are a few possible ways that can happen; priorities is the most likely. [17:16:04] Hm, I thought the local repo had highest priority, I will look again. [17:16:22] Ahmad_Sammour: You don't need to give a reason for your account, really, just explain about your bot. [17:16:31] Merlissimo: That depends on the project; unlike simple tools, that requires some consideration and consensus of some opsen because projects are relatively expensive. [17:17:23] Merlissimo: Maybe the mwreview project would suit you? That's what I (and others) use for semi-disposable mw installs. [17:17:34] Merlissimo: In your case, it doesn't seem too bad in isolation but I'm wondering whether mwreview would be better. [17:17:40] Bah. Ninja'd [17:19:59] Coren, can I trouble you to look at my apt setup? The priorities look right to me... [17:20:13] i only need a small mw instance which content i can clear easily for a new import test [17:20:20] same instance, proxy-abogott-7 [17:20:27] And the test is apt-get install python-flask [17:20:47] andrewbogott: Sure. Lemme go look; I'll install aptitude for this because it's generally a bit more verbose about why it chooses things though. [17:20:51] Correct behavior is to install version 0.10 vs. 0.8 which it insists on [17:21:17] Merlissimo, let me add you to mwreview and we'll see if that gets you what you want. [17:21:29] Merlissimo: do you want admin so you can create/destroy instances? [17:21:41] Coren, thanks [17:22:03] Hm, Merlissimo what's your wikitech username? [17:22:29] andrewbogott: i only need to be able to remove some wiki pages completly [17:22:34] merl [17:22:34] Good, now i will wait untill any admin acept my request (right)? [17:22:41] Ahmad_Sammour: right! [17:22:47] Or maybe you can get Coren to do it right now :) [17:24:00] Thanks a lot andrew [17:24:12] Actually, I need to go down that whole list, so yeah. [17:25:04] Ahmad_Sammour: {{done}} [17:25:48] Thanks coren! [17:25:55] Merlissimo: Ok, give me a few minutes and I'll set up an instance&wiki for you [17:26:10] thx andrewbogott [17:26:12] Then you'll have sudo on the wiki so should be able to do whatever you need. [17:39:31] Merlissimo: http://mwreview-merl.instance-proxy.wmflabs.org/wiki/Main_Page [17:39:38] Merlissimo: Can you log in and verify that you have sudo on that instance? [17:41:32] andrewbogott: seems to be ok: "merl@mwreview-merl:~$ sudo -l >User merl may run the following commands on this host: (root) NOPASSWD: ALL" [17:41:42] cool. [17:41:44] thx a lot [17:41:58] OK, so you should be all set -- maybe delete that project request for the sake of queue purity? [17:42:07] And, you bet! Let me know if you run into any trouble there. [17:42:31] It's cheap & easy to build you a new instance there, if you break that one :) [17:45:35] that's really all i need for my tests. thanks [18:27:30] need help! [18:29:52] I need help [18:29:52] Hi Ahmed__, just ask! There is no need to ask if you can ask [18:30:54] i have a problem in puplic keys and private keys. [18:31:21] Ahmed__: Can you be more specific? [18:32:16] I download putty generator and then i got puplic key and private key [18:32:45] but when i put the puplic key here: https://wikitech.wikimedia.org/wiki/Special:NovaKey [18:33:01] it said : "The provided SSH key was wrong or in unknown format. " [18:33:24] Ahmed__: Ah, sorry, I don't know much about windows. Are you already following the guide on wikitech? [18:33:56] maybe :) [18:34:27] But i dont understand much things [18:34:34] Ahmed__: I think that would be this: https://wikitech.wikimedia.org/wiki/Help:Putty [18:34:45] But otherwise… probably need to wait for another windows user to appear. Like maybe petan? [19:12:02] I cannot access tool labs, something changed? [19:14:16] lbenedix: can you be more specific? [19:15:01] Hm… Coren, did our recent fiddling break something? I can't connect to tools-login except as root [19:15:05] I tried to connect to tools-login.wmflabs.org [19:15:26] ... whu? [19:15:39] * andrewbogott does a puppet run to see what it says... [19:16:04] autofs seems sick. [19:16:04] i'm in tools-dev and I can access things [19:16:12] So no public keys. [19:16:14] okay, sshing to tools-login says dead pu [19:16:15] b [19:16:16] keys [19:16:23] * YuviPanda quits being useless [19:16:50] YuviPanda: https://gerrit.wikimedia.org/r/91541 https://gerrit.wikimedia.org/r/91620 [19:17:15] * Coren beats automount up. [19:17:19] "Server refused our key" [19:17:53] (03CR) 10Yuvipanda: [C: 032 V: 032] s/dynamic-proxy-api/dynamicproxy-api [labs/invisible-unicorn] - 10https://gerrit.wikimedia.org/r/91541 (owner: 10Andrew Bogott) [19:17:55] lbenedix: Yeah, the keys aren't there - either autofs is sick or gluster is. [19:18:09] okay [19:18:17] so just wait and try later? [19:18:19] andrewbogott: I'm not entirely sure if https://gerrit.wikimedia.org/r/#/c/91620/1/requirements.txt will work at all [19:18:38] autofs is a rickety pile of unstable digestion subproducts. [19:18:48] YuviPand, oh, you're right, that change is now moot. [19:18:50] * andrewbogott fixes [19:19:16] It seems to be stuck trying to mount gluster-exported stuff. [19:19:59] (03PS2) 10Andrew Bogott: Further attempts to make this work with setuptools [labs/invisible-unicorn] - 10https://gerrit.wikimedia.org/r/91620 [19:21:37] Coren, which gluster volume? I can restart it [19:21:58] Debugging autofs now. [19:25:16] Hi I want to connect to server by putty it refuse my public key. it worked till 3 hour ago! [19:25:26] what should I do? [19:25:30] wait [19:25:32] ;) [19:26:03] ok :) [19:28:49] reza1615: there's a known problem, Coren is working on it. [19:29:05] hmm thanks [19:29:40] Ah, the problem is actually older than this; automout maps were changed in puppet some time ago (and puppet couldn't cope with the change) but that didn't surface until the ssh thing was fixed since puppet never ran. [19:29:55] s/puppet couldn't cope/autofs couldn't cope/ [19:30:12] Because autofs craps its pants if it ever has to restart, apparently. [19:30:34] ima mount the filesystems by hand; next reboot will fix things automagically. [19:33:23] logging in should work again. [19:33:38] thanks a lot! [19:43:40] YuviPanda: proxy-abogott-8 is running a puppetized/serviceified unicorn. Want to have a look and verify that things look right to you? [19:43:58] andrewbogott: heading into a ~3 hour meeting, so not sure if I can :( [19:44:07] 3 hours yikes [19:44:11] andrewbogott: however, just point the test wikitech into it? [19:44:13] Anyway -- doesn't need to happen immediately. [19:44:15] andrewbogott: well, 3 hours, starting at 1:30 AM :P [19:44:30] YuviPanda: Also I updated https://gerrit.wikimedia.org/r/#/c/91620/ [19:45:18] okay [19:45:26] andrewbogott: will we be running the API directly? [19:45:37] andrewbogott: I suspect we'll want to put that in front of uwsgi or something [19:45:42] what do you mean, running directly? [19:45:50] andrewbogott: as in, just as 'python api.py' [19:46:00] that just starts the completely single threaded development server [19:46:10] Um… no idea, no opinion. Right now I have it as an upstart job [19:46:17] hmm, okay [19:46:28] should be okay to hook the upstart job to start it as uwsgi tho [19:47:02] hi! I have a question. I try to create file in my project: $ touch /data/project/chie-bot/xxx [19:47:06] touch: cannot touch `/data/project/chie-bot/xxx': Permission denied [19:47:34] linux says i'm in tool's group: $ groups [19:47:34] wikidev project-bastion project-tools local-chie-bot [19:47:43] and directory permissions are: [19:47:43] YuviPanda: I don't know what uwsgi is or what you mean. But the upstart script is… here https://gerrit.wikimedia.org/r/#/c/91499/ [19:47:49] drwxrwsr-x 6 local-chie-bot local-chie-bot 4096 Oct 23 19:55 chie-bot/ [19:47:59] where did I go wrong? [19:48:35] andrewbogott: ah, uwsgi is an 'application server'. Think of it as like apache2+mod_php, but for python. Currently it is just a 'toy' dev server that can't process more than one request at a time [19:51:35] YuviPanda: the redis and db stuff is shared among instances, right? So bringing up a new proxy and api on a new instance doesn't require anything to be copied over? [19:51:54] andrewbogott: not shared among instances, no [19:52:00] andrewbogott: needs copying, yes :D [19:52:02] Hm... [19:52:10] andrewbogott: as in, if we move it to a new machine, needs copying [19:52:12] I thought I pointed this to the new instance, but… I still seem the same list of proxies [19:52:17] which means I guess I didn't actually move it :( [19:52:32] perhaps the db moved with it? [19:52:33] not sure [19:52:36] db is just sqlite [19:52:37] Oh, unless the api has a hard coded ref to proxy-dammit [19:52:52] hmm? [19:52:53] no [19:52:56] it just talks to local redis [19:52:58] and local db [19:53:11] hm [19:53:47] hey guys, I'm having another problem with wikimetrics today. I couldn't ssh into the box yesterday so we rebooted it. We figured it was OOM and ignored it. But now that the problem is recurring, I'd like to find out why it's OOM [19:54:03] are there ways to diagnose what's going remotely? [19:54:03] Ah, memcache [19:59:20] Hey folks. I'm writing a guide to help researchers access WMF labs for data analysis. I'd like to actually walk through the process of setting up a test account to make sure that I have all of the steps right. Would anyone be willing to approve this test account for me and then delete it when I'm done. [19:59:35] halfak: hey! sure! [19:59:51] halfak: you mean labs or toollabs? [19:59:56] Both. :) [19:59:59] right [20:00:13] I'll sign up with "labs2_user_test" shortly. [20:00:18] ok [20:01:39] heh: "Account creation error: There was either an authentication database error or you are not allowed to update your external account." [20:02:07] andrewbogott: ^ [20:03:50] "A shell account name is required. It must start with a-z, and can only contain lowercase a-z, 0-9, and - characters." [20:04:45] Ahhh. I used underscores. [20:04:45] halfak: ^ [20:04:46] ah [20:04:51] * YuviPanda stops playing telephone [20:05:12] OK. "labs2-test-user" [20:06:07] looking [20:06:44] halfak: done [20:06:52] well [20:06:53] wait [20:06:59] i hit save, hasn't returned yet [20:07:06] halfak: done [20:08:11] Thanks! I'll go through the motions and then submit a request to tool labs. Can you help me there too? [20:08:34] halfak: ok [20:08:36] halfak: yeah [20:09:59] YuviPanda: This might take a little bit because I'm writing notes as I go. I'll ping you when I'm at the next place. [20:10:13] halfak: okay! [20:10:21] halfak: I've a meeting starting in 20mins and on for 3 hours, so I'll be about :D [20:10:43] Ok. No worries. If this doesn't get done today, it will be no trouble. [20:10:58] Thanks for your help. :) [20:11:29] halfak: heh :D [20:21:36] valhallasw, andrewbogott, Coren? could anyone help me please? [20:21:42] with my issue [20:22:43] Leloiandudu: What issue exactly? I'd be glad to help. [20:22:56] * Coren reads scrollback. [20:23:48] Leloiandudu: Did you log off and back in since creating the tool? [20:24:07] yes [20:25:23] I don't get it; the permissions are okay and you're in the group. Hm. There /is/ a delay before that propagates to the NFS server but it's only 180s at most. [20:25:44] (And should average 90) [20:26:01] yes... I created tool about 24hrs ago [20:35:52] I need help [20:35:52] Hi Ahmed__, just ask! There is no need to ask if you can ask [20:36:35] ok, while you are here, another question: i need to run my bot every 3 days. what's the best way to achieve this? would it be ok if I use cron for that? [20:38:00] i have a problem in access into winscp [20:39:53] any body can help? [20:41:15] Ahmed__: you can use the pagent.exe that comes with putty tools, load your key in that, then just use winscp without typing a pass and it should just work [20:42:14] maybe you'd have to enable using agent in winscp settings [20:45:48] Leloiandudu: Yep, cron is perfectly okay for that, though you want to use it to start your actual bot on the grid. [20:46:28] Coren: can you elaborate? where can i read about this? [20:47:04] Leloiandudu: https://wikitech.wikimedia.org/wiki/Nova_Resource:Tools/Help#Submitting.2C_managing_and_scheduling_jobs_on_the_grid [20:47:47] thanks, Coren! I failed to find this in help [20:47:53] Leloiandudu: as for you file write problem, I'm stumped. Every test *I* make works. Can you try writing files elsewhere in your tool's project directory to try and isolate the problem? [20:48:55] Leloiandudu: In particular, the "Submitting simple one-off jobs using 'jsub'" section is probably exactly what you need. [20:49:15] what if I create a directory under /tmp/ with my tool account? for testing purposes. and try to write there - will this suffice? [20:49:26] thanks [20:51:34] I created /tmp/chie and done chmod g+w [20:51:38] YuviPanda: Any chance you can approve my test application to Tool Labs quick? [20:51:42] halfak: yup [20:51:43] moment [20:51:51] now touch /tmp/chie/xxx works! [20:52:06] Please let me know if I forgot something so that I can add it to the instructions. [20:52:46] halfak: added [20:52:50] Thanks [20:52:52] yw [20:53:58] Coren: well.. you would'nt believe that. suddenly 'touch /data/project/chie-bot/xxx' works! i haven't done anything, i swear! i haven't logged of even [20:55:04] o_O [20:55:15] Dafu? [20:55:34] Not a chance your original tests were less than 180s after the group was created, was it? [20:55:50] no, i did them right before posting here [20:55:55] i'm very confused... [20:56:05] sorry for bothering you [20:57:08] YuviPanda: OK to delete that account now. [20:57:27] halfak: i'll leave that on a bit for you to experiment, will do that in a bit [20:57:33] halfak: also don't actually know if you can *delete* an account [20:57:38] will remove it from tools in a bit tho [20:58:19] Sounds fine to me. Thanks again for your help. :) [21:05:07] Leloiandudu: No, it was clearly a problem; it's even more annoying that it would just fix itself. [21:13:39] hey andrewbogott [21:13:59] Hey -- emailed so you can tend to meeting and sleep and such before responding :) [21:14:18] andrewbogott: yeah, saw that [21:14:31] andrewbogott: but my body has decided it's not going to sleep anytime soon, so investigating 500 [21:14:51] Aren't you in hour one of the three-hour meeting? [21:15:06] andrewbogott: yeah, but the parts I intensely care about just wrapped up [21:15:11] Ah, cool. [21:29:13] may we run interactive irc clients on tools? [21:31:02] morebots, what's up? [21:31:15] hm… labs-morebots ? [21:31:27] labs-morebots, are you broken? [21:31:27] I am a logbot running on tools-exec-01. [21:31:28] Messages are logged to wikitech.wikimedia.org/wiki/Server_Admin_Log. [21:31:28] To log a message, type !log . [21:31:30] Ah! [21:31:44] giftpflanze: Is that what you mean? [21:32:09] no, i dont't mean irc bots [21:32:33] hm, sorry, then I don't understand the question [21:33:01] i want to run irssi in a screen [21:33:47] Oh! Hm… I'd think not, but Coren has the last word. [21:34:04] giftpflanze: IIRC it is 'no' [21:34:15] giftpflanze: if it is a long running process, it should be running as a tool account in the grid [21:34:19] andrewbogott: "Zyxt" according to the OED. [21:34:26] is it somewhere written down? [21:34:34] Coren: ^ [21:35:44] giftpflanze: https://wikitech.wikimedia.org/wiki/Nova_Resource:Tools/Rules [21:36:20] aha! [21:37:20] so, that page doesn't say anything against running irssi in screen from a tool account? [21:37:29] Zyxt sounds like a word made up by an especially bold scrabble player [21:38:36] giftpflanze: sure, not *yet*, but it can! :P [21:39:05] a loop in the rules! i will use it as long as i can! ;p [21:39:28] * YuviPanda writes deamon that looks for and kills irssi everywhere! :P [21:39:46] oh noes [21:40:47] andrewbogott: so, I'm on abogott-8 [21:40:52] andrewbogott: where can i find api.py? [21:41:01] andrewbogott: path? [21:41:07] hm, seems i cannot run screen from my tool account [21:41:16] It's launched via /usr/bin/dynamicproxy-api [21:41:17] giftpflanze: indeed, no tty [21:41:20] andrewbogott: looking [21:41:28] but I think the source is in… /usr/lib someplace [21:41:51] was there a way to make it running? [21:42:16] andrewbogott: how can I kill it? is there a servie? [21:42:23] yep, service of the same name [21:42:36] ok [21:42:49] managed by… /etc/init/dynamicproxy-api.conf I think [21:43:57] andrewbogott: aaah, got it [21:44:11] /usr/lib/python2.7/dist-packages/invisible_unicorn [21:44:13] andrewbogott: the sqlite file needs to be created first time, so there's a first time run script that needs to be run [21:44:15] Is I think where the code lives [21:44:23] Ah, ok, I figured :) [21:44:33] giftpflanze: "Likewise, any process meant to execute without direct interaction should be submitted to the grid" should have been clear enough, really. If you want a bouncer, I'd recommend sending znc to the grid instead. [21:45:40] I'm not sure how generally comfortable I am with people running IRC bouncers on tool labs, to be honest, but that at least won't clutter the interactive boxen. [21:45:56] oh, missed that part … [21:46:35] znc is not installed ;) [21:46:54] andrewbogott: okay, live hack'd to not 500 [21:46:58] andrewbogott: how do we do those kinda things? [21:47:06] andrewbogott: also should we put the db in a particular place? [21:47:34] giftpflanze: Open a bugzilla, I'll probably get to it by tomorrow. [21:48:03] i don't like that process [21:48:56] giftpflanze: you can also submit the patch yourself, it is rather trivial... [21:49:15] Coren: I don't like that either (tools as personal bouncer) [21:50:07] remember, tool labs is toolserver replacement, where that was possible ;) [21:50:31] giftpflanze: sure, doesn't mean I have to be comfortable with it :P [21:50:44] giftpflanze: and where it made sense. I'm not sure what you want a bouncer for anyways, you don't get to have incoming connections from the 'net. [21:51:13] i don't understand [21:51:22] I mean, I know there are IRC bots that use bouncers so that they can be stopped and restarted without loosing history, and that's entirely legitimate. [21:51:46] giftpflanze: If you run a bouncer on tool labs, you wouldn't be able to connect to it from /outside/ tool labs. [21:52:07] that's no problem for me [21:52:12] YuviPanda: the db should probably be in /mnt [21:52:23]