[00:00:30] That form asks you for your preferred shell name. Are you entering one? [00:00:45] yes, same as the username. [00:05:31] Guest40473, well, on the plus side it looks like we don't already have a cipher. [00:05:44] Please try yet again, using your first choice for both account name and shell name? [00:05:55] ok. [00:07:15] just tried 2 times, and failed. [00:08:16] try again with all-lowercase shell name [00:08:24] ok [00:08:32] And don't do it twice, please, that confuses me :) [00:09:19] oh, that worked! all lower cases for shell name. [00:10:06] yep. I'll see if I can make the caption on that form a bit clearer [00:10:18] Thanks a lot Andrew :) [00:10:29] that would probably be a good idea. [07:36:43] !log stats setup a virtualenv and cloned the github repo to ~/python-cube-api. also installed libyaml-dev [07:36:44] Logged the message, Master [08:18:05] legoktm: what is the labs project? [09:16:16] YuviPanda some idea why ?status doesn't work on beta [09:16:52] hmm, not sure [09:16:57] dunno how it runs in tools itself [09:17:11] it runs same [09:17:15] both projects are identical [09:17:27] difference is that it works on one :P [09:17:30] heh :) [09:17:59] petan: the Redis role is +2'd in puppet :) [09:18:25] the dependent commit needs +2ing from someone, then we can add it to tools-mc, get rid of our homemade instance, and document it [09:18:25] cool, where is it on -beta? [09:18:32] no [09:18:42] petan: it's on tools-mc [09:18:43] we first need to set it up on beta [09:18:43] err [09:18:45] toolsbeta-mc [09:18:47] petan: it is setup on beta [09:18:49] right now [09:18:52] let's do things properly [09:18:53] ok [09:18:58] petan: so I tested it on toolsbeta-mc [09:19:01] with andrewbogott_afk's help [09:19:01] is it using that puppet class? [09:19:03] before it got merged [09:19:04] petan: yup [09:19:08] good [09:19:31] :D [09:19:48] petan: next would be to puppetize our 'restricted' verison of memcached... somehow [09:20:30] petan: is everything on tools puppetized? [09:20:36] like, installation of packages for people, etc? [09:21:27] not everything [09:21:31] packages should be [09:21:50] webservers for example are not [09:21:55] hmm [09:21:59] https://gerrit.wikimedia.org/r/#/c/70212/ is the patch, btw [09:22:03] ok [09:22:13] https://gerrit.wikimedia.org/r/#/c/70212/ isn't +2'd yet, since that'll affect production too [09:22:21] but will be [09:22:21] how many people use reddis now? do I need to announce a downtime? [09:22:31] petan: so far it's just me and addsleep [09:22:43] petan: we can't add it yet, the second patch needs +2ing [09:22:47] ok, he should wake up soon :D [09:22:49] we can do that hopefully tomorrow [09:22:50] k [09:23:00] petan: yeah, and even for him I patched his code to add support for redis ;) [09:23:06] petan: not many people know we had redis [09:23:06] so [09:23:25] we should keep that patched memcache somewhere [09:23:37] right now it's only on tools-mc:/mnt/share [09:23:40] petan: do we have icinga monitoring for stuff on toollabs? [09:23:45] no [09:23:55] ah, ok [09:23:55] depends [09:23:58] what you mean by "stuff" [09:24:01] servers are in icinga [09:24:05] tools not of coures [09:24:08] * course [09:24:10] no no, I mean [09:24:12] the grid engine [09:24:16] nope [09:24:16] or the apaches [09:24:22] there is bug for it [09:24:24] are the apaches on icinga? [09:24:24] ah [09:24:25] ok [09:24:27] apaches could be [09:24:28] dunno [09:25:01] okay, redis isn't [09:25:54] petan: don't think we'll ever use memcachedsasl. Agreed? [09:38:04] idk [09:38:14] I think memcache is quite a cool thing, it just needs some improvements [09:38:18] oh sure [09:38:26] i'm talking about memcachedsasl in particular [09:38:28] and right now lot of people use it and need it [09:38:30] sasl doesn't give us any advantages [09:38:33] sasl version not [09:38:40] yeah, so i was tlaking specifically baout that [09:38:41] it's not even supported much [09:38:44] we definitely need memcaceed [09:38:53] k [12:28:04] phpunit seems to be broken [12:28:32] on tools-login [12:32:55] Coren|DayOff, petan: is possible to have php-mcrypt installed? [12:33:07] hi [12:33:10] sure [12:33:19] you need it on webserver or execution nodes? [12:33:30] petan: On webserver :) [12:33:33] ok [12:33:52] petan: thanks a lot :) [12:33:53] at some point if cli uses it as well, we should probably put it to exec nodes as wel [12:34:13] petan: it may be a good point [12:34:52] !log tools petrb: installing php5-mcrypt on exec and web [12:34:55] Logged the message, Master [12:35:23] ew [12:35:34] php5-mcrypt : Depends: phpapi-20100525 [12:35:35] Depends: php5-common (= 5.4.15-1) but 5.3.10-1ubuntu3.6+wmf1 is to be installed [12:35:57] :\ [12:36:19] petan: weird. Ubuntu (only stock repo) has mcrypt compiled with the right php version [12:36:45] indeed that works [12:37:13] are you using a custom php version? [12:37:25] I don't know what Coren installed :) [12:37:29] :D [12:41:00] installed [12:41:41] paravoid, mutante: how do I tell puppet to install some version of package? [12:41:46] like apt-get package=version [12:41:51] * install [12:41:59] is it = in puppet too? [12:42:26] can I take a look at BracketBot's code? [12:42:42] petan: use the package class? [12:42:48] and setup dependencies appropriately? [12:42:55] YuviPanda what you mean [12:42:56] does A930913 have a public repo? [12:44:03] YuviPanda I have: package { [ [12:44:08] 'a', [12:44:10] 'b', [12:44:12] etc [12:44:16] yeah, that's how you do it [12:44:18] how I specifiy version there [12:44:24] 'a=version', [12:44:25] ? [12:45:15] hmm, reading docs at http://docs.puppetlabs.com/references/latest/type.html#package [12:46:09] I think you've to set it as the nesure property [12:46:10] *ensure [12:46:21] though it is not reccomended [12:54:03] no I didn't, Coren did :P [12:54:08] I don't even know what ensure is for [13:00:32] thanks petan :) [13:02:44] New review: coren; "It seems clear to me that the model is logically chown and 'chown foo *' changes the owner of direct..." [labs/toollabs] (master) - https://gerrit.wikimedia.org/r/70059 [13:02:55] 'nick coren [13:14:48] Coren: It turns out blocking any parts of 10.0.0.0/8 is not such a good idea, because it also blocks if such IPs show up anywhere in the X-Forwarded-For headers. So some random ISP's proxy includes XFF for their internal use of 10.x.x.x, and boom. [13:15:16] * YuviPanda nudges Coren with https://gerrit.wikimedia.org/r/#/c/70064/ [13:16:19] anomie: Ohcrap. We've been bugging the devs for XFF blocking in forever, and now that we got it it circles back and bites us. :-) [13:18:13] YuviPanda: Will merge as soon as Jenkins gets around to giving it its +2 [13:18:23] that doesn't have a +2? [13:18:24] * YuviPanda looks [13:18:43] Coren: err, why did you give it a -2? [13:19:01] Coren: there is a +2 from Jenkins [13:19:13] Because I'm a moron is why. :-) [13:19:27] "Case of the Mondays, eh?" :P [13:19:55] w00t [13:20:07] Yep. Plus a [bleep] sunburn. I should not have let my family take be outside; there's a danegrously close star out there bathing the place in radiation. [13:20:21] heh. 'Use protection' [13:20:33] does https://gerrit.wikimedia.org/r/#/c/70212/ get automatically merged now? [13:20:34] has a +2 [13:20:35] .. [13:20:36] I use very good protection. It's called "roof" [13:20:53] Coren: If we want to be able to make these blocks, I guess the thing to do is add another block option so individual blocks can be excluded from the XFF check. [13:21:23] Coren: <3 wonderful :) [13:21:34] anomie: Or restrict the blocks to the individual IPs we know bots are likely to run on. [13:21:37] thanks for the merge [13:21:51] YuviPanda: np [13:22:07] Coren: is all of tools labs puppetized? [13:23:07] Coren: also can merge https://gerrit.wikimedia.org/r/#/c/70103/? [13:23:17] we aren't going to use sasl for memcached ever, me and petan both agree on that [13:23:27] we'll need to puppetize the memcached instance, but it's not sasl [13:23:27] so [13:23:30] YuviPanda: Most of it except the webservers. [13:23:50] Coren: ah, okay. will they be puppetized at some point? [13:24:25] It's #4 on my todo [13:24:52] heh :) [13:27:05] anomie: Arguably anothing in RFC1918 should lead to XFF block. [13:39:23] hi Coren :) [13:39:35] Hello, fale [13:39:42] I can see an access.log file, but not an error.log. Where can I find it? [13:41:03] Coren: hi, a question. how can we create a wiki (or something so much better moving a wiki from tools) in beta cluster? [13:41:51] Amir1: I... don't know! [13:42:06] fale: Is your code PHP? [13:42:19] Coren: yep, and returning 500 [13:42:32] Coren: who know? [13:42:34] fale: it's probably permissions [13:42:39] petan: ^ [13:42:58] petan: have to be world-readable? [13:43:23] addshore, ping [13:43:23] fale needs to be a) owned by your tool account, b) must not be world writable c) must be readable by www-data [13:43:33] fale: ... You /should/ have a php_error.log with errors starting up PHP [13:43:45] Coren: that doesn't display permissions errors :/ [13:44:01] fale: If you don't, then it never got as far as trying to start it, so it means your script isn't exectuable and/or isn't owned by your tool. [13:44:04] Coren, S7 is still broen. [13:44:05] Coren: in /data/project/lists there is not [13:44:13] Cyberpower678: What's broken about it? [13:44:22] Amir1 why would you ever move a wiki from tools to betacluster o.O [13:44:28] that is far worse place to host a wiki [13:44:29] petan: thanks, going to try the permission fix :) [13:44:30] Coren, everything. [13:45:09] Coren, [13:45:10] cyberpower678@tools-login:~$ sql metawiki [13:45:10] ERROR 2003 (HY000): Can't connect to MySQL server on 'metawiki.labsdb' (110) [13:45:10] Make sure to ask for a db in format of _p [13:45:34] Hm. Was that alias forgotten? Lemme check. [13:46:45] petan: we (me and another Amir) made a universal wiki for testing mediawiki features for RTL langs [13:46:47] http://blog.wikimedia.org/2013/05/30/test-features-in-a-right-to-left-language-environment/ [13:46:59] Cyberpower678 metawiki works to me (alias) db is down I guess [13:47:07] but now We can't run Parsoid on it [13:47:08] ah right [13:47:13] I see [13:47:20] petan: 775 seemed good to me, but the 500 is still there. And I'm not allowed to change owner to make it local-lists:www-data [13:47:25] Amir1 why not? [13:47:28] Amir1: i think best bet would be to create a new project on labs and use it there. [13:47:36] fale you can use take utility [13:47:40] Amir1: that way you get an instance + root, so you can do whate you want :) [13:47:46] parsoid should be simple to run there [13:48:02] fale you don't need to set group to www-data it just needs to be readable by www-data [13:48:09] Coren: Ah, it seems it was. Give me a minute. [13:48:10] fale: if it's world readable then it's ok [13:48:20] * YuviPanda douses Coren in ice cream [13:48:22] petan: no man for take :( [13:48:25] talking to yourself again, are you [13:48:36] fale: not on production, because my package is waiting for approval... [13:48:36] Coren, you just pinged yourself. :p [13:48:40] petan: because whenever i ran it in tools, VE and Parsoid can't connect because both must be ran in a same host which is not possible [13:48:51] fale: just type take "filename" [13:48:55] and pray [13:49:01] hey Cyberpower678 ! [13:49:44] Amir1 ok I think you should ask for a new project on labs maybe... I can't really help here, but beta is not a place for it either [13:49:48] YuviPanda: I think I'm not authorized to do this, [13:49:52] petan: no error returned, but not fixed either [13:49:59] fale ok sec [13:50:04] petan: I requested a billion years ago [13:50:05] Amir1: yes, but you can ask someone who is to create it for you. [13:50:12] Amir1: usually Ryan_Lane creates it for me [13:50:19] but he's not around [13:50:20] Amir1 billion years ago, there was no wikimedia :P [13:50:20] addshore, did you look at the updates? [13:50:22] http://deployment.wikimedia.beta.wmflabs.org/wiki/Global_Requests#fa.wikipedia_and_he.wikipedia [13:50:39] Amir1: betalabs is the wrong place for that, I think. [13:50:39] Coren any chance my user-friendly take get reviewed? [13:50:41] labs != betalabs [13:50:59] YuviPanda I already told him [13:51:27] fale: what is name of tool [13:51:34] petan: lists [13:51:42] petan: o_O It has been. A lot. By at least four people? Or do you mean transcribe those comments to the actual gerrit pacth? If the latter, I can do so sometime early this afternoon. [13:51:50] oh ok [13:52:00] fale: option MultiViews not allowed here [13:52:06] fale: that is error I see in log [13:52:11] Amir1: so I suggest waiting for Ryan_Lane to turn up, ask him for a project, and he'll be able to create it for you :) [13:52:22] petan: I see [13:52:22] I am sure other people can too, but I do not know who they are :( [13:52:25] And I'm not one of them [13:52:43] Coren: ok and what is a result of that review? did you find any issue? [13:53:03] so I've two options, run the wiki on an instance (as YuviPanda said) or request for new project in labs, [13:53:10] all the reported issues so far were already fixed, many days ago [13:53:13] they're both the same, Amir1 :) [13:53:13] petan: There are plenty. It's still completely unsecure by default at the last version I checked. [13:53:24] Amir1: so you 1. get a project, 2. create an instance in the project and run it inside :) [13:53:24] Coren: what makes it unsecure? [13:53:34] Amir1: I think there's a request queue for projects on wikitech. [13:53:38] ok [13:53:49] petan: You're still using pathnames, and have dozens of race condition exploits. [13:54:18] Coren where? I am using pathname on 2 places, in debug log and in function which opens the filedescriptor, no race conditions I could see anywhere [13:54:34] scfc_de: how many wikis are in this queue? [13:54:37] petan: Perhaps you've updated it since, I'll have to check. [13:54:44] Amir1: https://wikitech.wikimedia.org/wiki/Special:FormEdit/New_Project_Request, though nagging Ryan_Lane is probably faster :-). [13:54:45] Coren: that's possible [13:54:56] Coren, what's the status of S7 [13:55:11] Amir1: https://wikitech.wikimedia.org/wiki/Special:Ask/-5B-5BCategory:New-20Project-20Requests-5D-5D-5B-5BIs-20Completed::No-5D-5D/-3FProject-20Name/-3FProject-20Justification/-3FModification-20date/format%3Dbroadtable/sort%3DModification-20date/order%3Dasc/headers%3Dshow/searchlabel%3DOutstanding-20Requests/default%3D(No-20outstanding-20requests)/offset%3D0 [13:55:11] !s7 is dead [13:55:11] Key was added [13:55:15] Cyberpower678: I'm trying to figure out why it work directly, but not through SQL [13:55:17] Amir1: plus https://wikitech.wikimedia.org/wiki/Help:Getting_Started has information on projects vs instances, etc [13:55:18] !s7 | Cyberpower678 [13:55:19] Cyberpower678: dead [13:55:23] :P [13:55:48] !s7 is S7 is dead [13:55:48] This key already exist - remove it, if you want to change it [13:55:59] !remove s7 [13:56:05] !doc [13:56:06] petan: where are the log file you looked at? [13:56:06] There are multiple keys, refine your input: docs, documentation, [13:56:15] !remove is if you want to delete a key, try !key del [13:56:15] Key was added [13:56:30] !key del s7 [13:56:30] Unable to find the specified key in db [13:56:35] !s7 del [13:56:36] Successfully removed s7 [13:56:36] fale: on webserver, not readable :/ [13:56:44] a question: is there a way to move database of this wiki (old wiki on tools) to the instance? [13:56:50] Aha [13:56:53] I don't want waste edits of people [13:56:57] Cyberpower678: they look good :) [13:56:58] Amir1 yes [13:56:59] Amir1: you should be able to dump the sql and load it up [13:57:04] I still havn't had much of a chance to play yet :/ [13:57:08] Amir1 you can use mysqldump [13:57:11] petan: :D. I should have fixed the multiview stuff, but sill 500 [13:57:27] [Tue Jun 25 13:55:31 2013] [alert] [client 10.4.1.89] /data/project/lists/public_html/.htaccess: Option MultiViews not allowed here [13:57:34] very good [13:57:35] addshore, you should download the new updates from GitHub. Then you won't ever have to dowload Peachy again. [13:57:46] and URL will be same? [13:58:06] petan: weird, I just added a conditional statement on that line [13:58:23] what about just removing that entirely [13:58:26] Amir1: URL will be different, but with a little PHP you can setup redirects from current tool to your new instance [13:58:27] (simply we can redirect it, never mind, stupid question) [13:58:31] petan: yep, going to [13:59:29] petan: 500 went away... a good 404 is waiting on the line :D. Thanks :) [13:59:49] hmm [13:59:50] :o [14:00:38] petan: I think I'll try some simple php script before trying again with a framework [14:00:41] Cyberpower678: Should be fixed from -login now. :-) [14:00:53] Cyberpower678: Not allowing -login in was... a bit silly of me. :-) [14:02:29] fale: Does it work for you now? [14:03:00] So i wait here until Ryan_lane shows up [14:03:28] Coren, so how do I connect to centralauth? [14:04:15] Cyberpower678: Ah, centralauth needs its views still (the standard wiki ones clearly won't work). Sometime this afternoon for that one. [14:04:43] Coren: nope, is returning a 404 error from the main tools website (http://tools.wmflabs.org/lists/) [14:05:49] the problem is my .htaccess. Going to fix it [14:05:52] fale: I see "Welcome to Lists Project homepage!" [14:06:11] Coren: yep, just removed the routing lines from htaccess :D [14:08:48] Coren, any news from legal? It's been weeks since you said it's going to be weeks. :p [14:11:06] Coren, petan: I really messed up the htaccess -.-. Fied now :) thanks to you both [14:14:03] YuviPanda: In modules/toollabs/manifests/redis.pp, the top line says Class: toollabs::*execnode* :-). [14:14:14] copypaste faile [14:14:45] let me fix [14:16:58] scfc_de: Coren https://gerrit.wikimedia.org/r/70422 [14:20:37] addshore, https://tools.wmflabs.org/xtools/pcount/index.php?name=Cyberpower678&lang=meta&wiki=wikimedia :p [14:21:00] To think all of these error messages were being suppressed on the toolserver edit counter. :p [14:21:01] :/ [14:21:07] HAHA! [14:21:12] I've got some work to do. :/ [14:21:24] at least they are only notices ;p [14:21:40] But they define the namespaces. [14:21:54] Namespaces the edit counter will not recognize. [14:34:29] hashar: Are you involved with the Gerrit -> Bugzilla notifications? [14:39:00] scfc_de: I am not [14:39:19] scfc_de: qchris and ^demon are. You can open bugs agains Wikimedia > git/gerrit (IIRC) [14:39:45] hashar: Merci. [14:40:59] hashar: Someone was faster: https://bugzilla.wikimedia.org/show_bug.cgi?id=46997 :-). [14:44:52] petan: around? :> [14:53:04] addshore, want to work on the Peachy wiki? [14:53:15] Provide valuable documentatio. [15:03:02] but I havn't started using it yet :D [15:05:39] addshore, busy, busy addshore [15:05:43] :) [15:06:23] addshore, I almost know Peachy inside and out. If you need help, let me know. [15:06:40] Hm. The gridengine accounting file will also need to be logrotated or something. [15:07:04] Coren, you haven't answered my question yet. ;) [15:07:18] Cyberpower678: What question? [15:07:36] Coren, any news from legal? It's been weeks since you said it's going to be weeks. :p [15:08:23] Cyberpower678: News from legal? No. And not for a while, I expect, given their current focus on the privacy policy (which they also probably consider to be a dependency on the answer - whether something is okay to share given the privacy policy rather depends on what the privacy policy /is/) [15:09:07] Ok. [15:09:22] Coren, can you MS me once there's news? [15:09:27] It /does/ make sense when you think about it. [15:09:33] Cyberpower678: You'll be the first to know. :-) [15:09:45] MS=MemoServ for clarification. [15:11:36] Coren, thanks. [15:12:43] Coren: On Toolserver, the first entry in /sge/GE/default/common/accounting seems to be from May 21st, and the size is 107 MByte. Don't know if it's rotated by standard logrotate or something inside SGE, though. [15:13:23] * Coren will look it up. [15:13:26] Ah: http://arc.liv.ac.uk/SGE/howto/rotatelogs.html [15:14:12] Ah. "Best to use logrotate". [15:17:54] Well, that's probably something to test on the actual master and not to try to reproduce on a puppet mockup :-). [15:42:54] Coren: is getting root on toollabs essentially 'keep doing infrastructure work, and if you have to ask for root you do not need it'? :) [15:43:31] YuviPanda: I suppose it is. :-) There are no hard rules. [15:43:53] well, okay then :) [15:44:12] * YuviPanda puts 'get root on toollabs' in 'goals' list [15:44:46] YuviPanda: I do intend to keep the total number relatively low, though. It doesn't take that many sysadmins to make coordination problems overtake the advantage of numbers and better coverage. [15:44:58] Coren: sure, sure. common sense, etc [15:45:24] hell, if there are enough admins then I don't need to be one :) [15:46:55] Coren: can you add role::labs::tools::redis to tools-mc? [15:47:10] wtf [15:47:11] I joined and half of freenode died? [15:47:16] Coren: I tested it on toolsbeta-mc earlier [15:47:18] petan: ^ [15:47:25] petan: Stop breaking freedone. [15:47:27] :-) [15:47:30] I'll add docs later today, and email out to labs-l :) [15:48:32] !ping [15:48:32] pong [15:48:33] hmm, or maybe we need to uninstall the hand-installed version of redis first? [15:53:26] !ping [15:53:26] pong [16:45:08] Centralauth now up. [16:45:20] * Coren remains amused at his low gu_id [16:55:16] Coren, do you happen to know if there's an apt package that installs Moose? I'm trying to run some third-party code, getting Can't locate Moose.pm in @INC [16:55:22] And want to avoid cpan... [16:56:04] andrewbogott: I expect there is given the presence of libany-moose-perl [16:56:26] hm, installed that already to no avail [16:56:52] Yeah, but that uses 'moose or mouse' whichever is available; I expect there is a way to make both available then [16:57:17] p libmoosex-app-cmd-perl - Perl module combining App::Cmd and MooseX::Getopt [16:57:28] That should include Moose in its dependencies. [16:57:50] Watching what it wants to install might be informative. [16:58:28] libmoose-perl [16:58:40] That was too simple. :-) [16:58:44] Coren, yep, that did it, thanks. [17:45:33] Coren did you find the issues in code? [17:46:07] petan: I haven't gotten around to review it yet. It's still early afternoon. :-) I'm going some ops stuff atm, will do so soon. [17:46:25] btw I found issues in your code [17:47:08] it cant handle multiple folder arguments but it is fixed in my version... [17:58:59] petan: Test case? [18:00:56] scfc_de? [18:01:13] I thought you already saw that bug [18:01:42] But Coren fixed that already? [18:02:12] * anomie reads backscroll [18:04:47]