[01:28:45] anyone still up and could figure out why puppet isn't working on an instance? [01:37:57] Thehelpfulone: What's the error message? [01:38:51] JohnLewis: ^ [01:38:59] I think it's just when we're trying to create the instance scfc_de [01:39:20] I've not look for a specific error message. Puppet just failed off creating an instance [01:41:54] JohnLewis: What do you mean by that? [01:43:29] scfc_de: created an instance and puppet failed. [01:48:39] JohnLewis: I'm too tired for this. Good night! [01:48:51] yeah same here [01:51:01] bye [04:08:15] 3Wikimedia Labs / 3Infrastructure: Increase quota on Labs for Language Engineering team - 10https://bugzilla.wikimedia.org/69770 (10Kartik Mistry) 3NEW p:3Unprio s:3normal a:3None We at Language Engineering team, are using and going to use Labs for various experimental features. Current quota (due to... [06:36:37] "Host '10.68.17.174' is blocked because of many connection errors; unblock with 'mysqladmin flush-hosts'" [06:36:38] I er uhh [06:37:52] :o [08:09:18] anyone here who's able to talk about wether my software would be allowed and how it can be achieved on labs? [08:36:42] 3Wikimedia Labs / 3Infrastructure: Increase quota on Labs for Language Engineering team - 10https://bugzilla.wikimedia.org/69770#c1 (10Antoine "hashar" Musso) Current quota: * Cores: 19/20 * RAM: 38912/51200 * Floating IPs: 0/0 * Instances: 13/20 * Security Groups: 0/10 So the project is CPU starved. You... [08:48:41] springle / Coren: dewiki replication on labsdb1002 has stopped more than 15 hours ago [09:11:43] hashar: in some experiment (apertium) we really need CPUs :) [09:12:18] I can imagine :] [09:15:59] 3Wikimedia Labs / 3(other): (Tracking) Database replication services - 10https://bugzilla.wikimedia.org/48930 (10merl) [09:16:01] 3Wikimedia Labs / 3tools: dewiki / wikidatawiki replication (s5) has stopped on labsdb1001/labsdb1002 - 10https://bugzilla.wikimedia.org/69776 (10merl) 3NEW p:3Unprio s:3blocke a:3Marc A. Pelletier No new data since 20140819190634 on dewiki and wikidatawiki databases available. [10:43:58] jongleur: hey [10:44:13] uihhh ;) someone who answers ;) [10:44:15] hi YuviPanda|zzzz [10:44:26] (although a sleeping one obviously ;) ) [10:44:34] Just woke up [10:45:16] jongleur: what is your tool? [10:45:55] Look-and-listen-map [10:46:11] not yet on labs, but was on toolserver before [10:46:13] Aaah [10:46:17] I saw the email [10:46:19] So [10:46:27] Toollabs has a tomcat service [10:46:27] yes [10:46:31] And you can use that [10:46:37] We also have osm db [10:46:44] So you can use that too [10:46:52] And I'm pretty sure it is allowed [10:46:53] tomcat sounds useful. OSM db means osm-mapnik, right? [10:47:06] Only restriction being that your tool must be open source [10:47:10] I believe so [10:47:34] Osm accounts aren't automatic yet - you need to request them but they are usually granted quickly [10:47:56] the most current stuff about osm I found is this https://wikitech.wikimedia.org/wiki/Openstreetmap_Databases and it tells it's out of date ;) [10:48:43] jongleur: yeaaah it was mostly fixed last week but not rolled out into full access yet. [10:49:05] I'll ask around to make sure but it should be fine? [10:49:07] great to hear. I stopped following the discussions around march or something like that due to my thesis [10:49:14] Right [10:49:28] !help [10:49:28] !documentation for labs !wm-bot for bot [10:49:35] !documentation [10:49:48] Brr [10:52:07] okay... so I should create an account, ask for getting access to osm - whatever variant of osm database is available then (the wiki page only lists possible ones, but as it's out of date and there's no detail I don't know what's available then [11:01:14] jongleur: right [11:01:21] jongleur: https://wikitech.wikimedia.org/wiki/Nova_Resource:Tools/Help [11:02:44] YuviPanda: thanks for your help. Is labs-l the right way in general to get more detailled help? this channel seems to be less helpful most of the time (no offense, but usually I wait for several hours til days, if I get any response at all, so asynchronous communication like mail seems to be more promising) [11:03:02] jongleur: indeed, labs-l is best for async communication [11:03:18] jongleur: and this channel is also usually more active during US / EU daytimes [11:03:22] (usually) [11:03:35] jongleur: although the particular combination of stuff you'll be using (tomcat + osm), not many people do [11:03:42] ;) everytime, but when I want to know something :P [11:03:47] I know [11:05:28] YuviPanda: a last question, as I'm not sure I understand it completely... labs is for tools, but labs-beta isn't for tools in beta stadium but for puppet configurations in beta, right? [11:05:59] jongleur: aaah, so there are a bunch of 'labs' [11:07:05] jongleur: 1. toollabs, which is tools.wmflabs.org, which is like toolserver, 2. 'betalabs', aka 'deployment-prep' which is for a production-like environment for people to test mediawiki changes on / run browsertests 3. 'labs' which is just an OpenStack environment providing VMs to whoever is interested for whatever purpose - from testing puppet changes to [11:07:05] running test mw instances to building servies like quarry.wmflabs.org [11:07:57] ah, yes - sorry... stumbled over not labs being the replacement for toolserver, but tool- being the replacement part ;) [11:08:08] question answered, thanks [11:09:26] jongleur: :D [14:18:30] 3Wikimedia Labs / 3Infrastructure: Increase quota on Labs for Language Engineering team - 10https://bugzilla.wikimedia.org/69770#c2 (10Andrew Bogott) 5NEW>3RESO/FIX I've raised your quota for Cores. Do try to free any instances you can live without though... labs is getting pretty busy! -A [14:34:36] YuviPanda: hello there! :-] [14:34:51] YuviPanda: do you have any frontend for the diamond collected stats ? :] [14:35:11] I know there is a graphite instance somewhere, was wondering whether you have set up some kind of nice frontend on top of it [14:36:44] hashar: not yet. we used to have graphite.wmflabs.org but that's dead (not enough resources). we now have labmon1001, which is on real hardware and about to be completed soon [14:36:56] \O/ [14:36:59] hashar: and then I'll have something akin to tools.wmflabs.org/giraffe/index.html#dashboard=ToolLabs+Basics&timeFrame=1h [14:37:08] yeah giraffe seems nice [14:37:13] since it 100% client side [14:37:53] YuviPanda: the beta cluster can definitely use some kind of dashboard :] [14:38:16] hashar: yeah, will get it done :) awaiting some network config for labs instances to be able to talk to labmon [14:38:31] hashar: even better, when that's done we can have graphite based icinga checks for deployment-prep [14:38:49] awesome [14:39:08] http://graphite.readthedocs.org/en/latest/tools.html has a bunch of tool [14:39:22] one of them point to https://github.com/douban/graph-index which is a django app [14:39:28] that pairs with diamond metrics [14:48:20] hashar: yeah, but graphite can have other non-diamond metrics too [14:48:59] YuviPanda: yeah true [14:49:12] YuviPanda: anyway, needs labsmon to be set up [14:49:12] hashar: so I guess I'll just have an autogenerated giraffe thing [14:49:18] hashar: true :) [14:49:31] hashar: hopefully someone can find time to do the network config this week [14:55:51] 3Wikimedia Labs / 3Infrastructure: Increase quota on Labs for Language Engineering team - 10https://bugzilla.wikimedia.org/69770#c3 (10Kartik Mistry) Thanks! We will delete unneeded instances as soon as experiments are over! [15:21:08] 3Wikimedia Labs / 3deployment-prep (beta): Add RecentActivityFeed extension to beta labs - 10https://bugzilla.wikimedia.org/69785 (10Sage Ross) 3NEW p:3Unprio s:3normal a:3None This extension is being built as a replacement for the activity feed within the EducationProgram extension, and is also expe... [15:45:52] 3Wikimedia Labs / 3deployment-prep (beta): Add RecentActivityFeed extension to beta labs - 10https://bugzilla.wikimedia.org/69785#c1 (10Chris McMahon) Do you have an estimate of when this would be ready to go to production? [16:13:50] 3Wikimedia Labs / 3deployment-prep (beta): Add RecentActivityFeed extension to beta labs - 10https://bugzilla.wikimedia.org/69785#c2 (10Sage Ross) @Chris McMahon: Andrew Green has started looking over the code and will be doing a formal review soon, I think. Functionally, it's in good shape from my perspecti... [16:26:36] 3Wikimedia Labs / 3deployment-prep (beta): Add RecentActivityFeed extension to beta labs - 10https://bugzilla.wikimedia.org/69785#c3 (10Chris McMahon) I ask because we have a loose policy of not installing new features on beta labs without a solid release date for the feature. We do this to reduce the amou... [16:30:50] 3Wikimedia Labs / 3deployment-prep (beta): Add RecentActivityFeed extension to beta labs - 10https://bugzilla.wikimedia.org/69785#c4 (10Sage Ross) Thanks, that's very helpful info. So we'll wait on the code review before putting it on beta labs, and in the meantime, I'll see about putting it on another labs... [16:34:05] 3Wikimedia Labs / 3deployment-prep (beta): Add RecentActivityFeed extension to beta labs - 10https://bugzilla.wikimedia.org/69785#c5 (10Chris McMahon) Thanks, beta labs should be very much like production. We can work with Greg and Antoine and Ops to get this to beta and on to prod. [16:54:11] Coren and/or scfc_de, will you create test accounts here, and then tell me the usernames you select? http://wikitech-test.wmflabs.org/w/index.php?title=Special:UserLogin&returnto=Main+Page&type=signup [16:55:58] andrewbogott: Created: Teßt [16:56:23] Really, an esset? You aren't making this easy :p [16:56:57] andrewbogott: /my/ keyboard has a compose key. Doesn't yours? :-) [16:57:23] andrewbogott: (shell account name is, predictably enough, 'tesst') [16:57:48] ok… can you log in here? horizon-test.wmflabs.org/horizon/ [16:58:00] be warned that auth fails about 30% of the time there for no good reason, so you might have to try a few times. [16:58:54] More like 100% atm [16:58:58] hm, lemme try mine [16:59:30] i can log in, so I must've messed up something. [16:59:40] My credentials do work again on the wikitech-test [16:59:47] So it's not that [17:00:02] Hm. Username is Wikitech account not shell account yes? [17:00:20] oh, good question. [17:00:35] I think so, but try the other [17:00:57] Wiki username -> "An error occurred authenticating. Please try again later." [17:01:08] Shell username -> "Unable to authenticate to any available projects." [17:01:23] interesting… so it's doing something! [17:01:26] Lemme add you to some projects. [17:03:15] Try now? [17:03:24] (both, I still don't know which name it's checking) [17:13:48] Coren, if that doesn't work I may need you to recreate with an ascii account name :( [17:15:20] andrewbogott: both give the same (distinct) errors. Lemme create an account in ascii but beware that we have users with lots of non-latin names. [17:15:28] * andrewbogott nods [17:15:45] Coren, one other thing -- if you enter the wrong password does the error message change? [17:16:02] andrewbogott: On account creation for "Tim Landscheidt (Test)"/"scfc-test", I get "Account creation error: There was either an authentication database error or you are not allowed to update your external account." Is the database shared with wikitech? [17:16:35] scfc_de: no, local db [17:16:47] but, actually... [17:16:58] andrewbogott: "Invalid user name or password." on bad password w/ shell account name. With Wiki name, no difference (still "An error occurred authenticating. Please try again later.") [17:17:08] I imported ldap from production. So if you have an account by that name on wikitech it'll conflict [17:17:30] "Tesst2" (wiki name) -> "You are not authorized for any projects." [17:17:37] Yeah, that 'an error occurred' makes me think it's freaking out about encoding. [17:17:49] ok, lemme set up Tesst2 and see if that works any better [17:17:58] "tesst2" (shell account) -> same [17:18:21] So it looks like it accepts either, but really doesn't like UTF-8 [17:18:50] ok, you should be in the 'testproject' project now. can you log in? [17:19:11] (Side note: On that error screen, the instance shell name isn't copied from the initial form.) [17:19:14] andrewbogott: Yes. [17:19:18] cool. [17:19:24] And you have a 'project' tab and that's it, yes? [17:19:33] andrewbogott: I do. [17:19:35] Lemme make you an admin so you can see everything [17:19:40] kk [17:20:06] ok, reload [17:20:14] (or you might have to log out/in, don't know how often it checks in w/keystone) [17:21:04] andrewbogott: Ah! The wiki has the central notice with that exact warning (do not use conflicting usernames, but that isn't shown on the account creation form :-). [17:21:33] andrewbogott: reload did nothing. [17:21:58] andrewbogott: Got an admin pane now [17:22:03] andrewbogott: Re mail, it said: "devwiki could not send your confirmation mail. Please check your email address for invalid characters. Mailer returned: Unknown error in PHP's mail() function." Very certain that my mail address is working and does not contain invalid characters :-). [17:22:10] ok, so had to out/in to get the admin pane? [17:22:19] * Coren nods. [17:22:27] scfc_de: yeah, most likely mail is totally broken on that box. [17:22:38] Anyway, what's your shell name? [17:22:52] 'tesst2' [17:23:03] I meant, scfc_de, what's your shell name [17:23:16] On wikitech-test [17:24:04] andrewbogott: scfc-test99 [17:24:23] Coren: So, if you're curious, I've made you cloudadmin on wikitech-test so you can see how changes correlate between the two interfaces. The underlying setup is a bit broken, though, so e.g. instances won't spin up properly. [17:24:33] And Horizon hates our ldap user system so won't let you manage users... [17:24:46] But you can at least see look and feel. Some good things and some bad things are immediately evident. [17:25:11] scfc_de: ah, sorry, I guess I need both. On-wiki name is? [17:25:12] andrewbogott: At first glance, it looks worth the effort to tweak it to our needs. [17:25:23] Anybody used with MediaWiki Vagrant VM? I want to test a deployment script from a wmf branch and I cannot do git updates from mediawiki/ directory [17:25:42] andrewbogott: "Tim Landscheidt (Test 99)". Are the links "List Instances"/"List Projects" in the side bar supposed to be working? [17:25:43] Coren: I'm tempted to agree. We'd have to make a list of differences, and decide whether it's worth reimplementing in Horizon or using some kind of hybrid setup. [17:26:01] * Coren nods. [17:26:18] Does Horizon provide an extensions scheme we can use? [17:26:29] Ah, http vs. https. [17:26:50] And "No such special page". [17:27:15] ok, scfc_de, you should be able to log in to Horizon as well. http://horizon-test.wmflabs.org/horizon [17:27:23] I gave you all the same rights as Coren in case you want to compare notes. [17:27:48] Coren: I've coded for Horizon a bit. I would say that it is well designed, very extensible, and only barely constructed. [17:28:08] So, adapting it for wikitech would be fun, but also a lot of time. [17:28:49] There's no real reason we couldn't have side-by-side interfaces while we work on it though. [17:28:53] Other than double the documentation [17:29:42] andrewbogott: Well, we can still direct most people at the wikitech thing until we are confident enough; just point the more savvy users at the horizon one. [17:29:46] The first glaring problem I saw is e.g. if you manage projects… the filter only filters the visible page, not the total list. [17:29:48] andrewbogott: Horizon = UI for OpenStack in the same way we use wikitech for project management ATM, or one step up? [17:29:59] As though they've never considered more than 20 tenants in a cloud. [17:30:55] scfc_de: It largely overlaps the OpenStackManager extension. So, some features are currently in one, some in the other. [17:31:22] Some of what OpenStackManager does is pretty specific to our setup, though, so we'd need to make some custom panels that would never make it upstream. [17:31:32] andrewbogott: Thankfully, fixes to /that/ limitation could be sent upstream. [17:31:46] Yeah, lots of what we need would be upstreamed. [17:33:58] Coren, so now that you have a valid login, can you reconfirm that it hates the esset? [17:34:11] And, scfc_de, can you verify that you're able to log in to horizon with both your wiki name and your shell name? [17:34:15] * Coren tries this. [17:34:24] Maybe I'm too superficial, but already the pie "charts" of resource usage appeal to me :-). On second look, "Hypervisors"' shows "VCPU Usage Used 5 of 4", yet the pie is only 3/4 full. Same for memory and disk. [17:34:54] scfc_de: yeah, I like the running-quota-tally. [17:35:01] Teßt -> "An error occurred authenticating. Please try again later." tesst -> "Unable to authenticate to any available projects" [17:35:30] andrewbogott: "Tim Landscheidt (Test 99)" => "Invalid user name or password."; shell name "scfc-test99" => successful. [17:35:43] ok, so it wants shell name. [17:36:05] I guess in the real world we'd probably authenticate with wikitech and then use a token. So maybe this is moot. [17:40:38] Coren, try tesst one last time? [17:43:11] andrewbogott: tesst works now. [17:43:44] great! Ok, so at least that isn't a dealbreaker. [17:43:44] But the login screen really dislikes Teßt [17:44:07] Yeah, hopefully we can cut that out entirely. [17:44:24] That's all I have for now. thanks y'all. [17:45:21] 3Wikimedia Labs / 3deployment-prep (beta): Add RecentActivityFeed extension to beta labs - 10https://bugzilla.wikimedia.org/69785#c6 (10Chris McMahon) Adding Chris Steipp because this will need security review at some point. [18:01:11] i recently created a new instance in a new project and i notice on puppet run that mounting /data/project fails with "No such file.." on server side. labstore issues? just wait a while? [18:03:34] mutante: Do you have "Create shared project storage" checked at Manage Projects => Configure? [18:04:15] scfc_de: yes, the box is checked. that was a default though [18:04:45] Execution of '/bin/mount /home' returned 32: mount.nfs: mounting labstore.svc.eqiad.wmnet:/project/sensu/home failed, reason given by server: No such file or directory [18:05:08] it's not created on labstore itself (yet?) [18:05:41] mutante: I think there's a daemon sitting there that creates them, but you need andrewbogott or Coren to debug. [18:06:55] scfc_de: ack, thx [18:10:28] mutante: there's a bit of a race. Typically rebooting your instance will resolve the problem. [18:11:07] andrewbogott: ok, doing so [18:17:09] do people install puppet modules on labs using "puppet module install" ? [18:19:15] i tried just to try and [18:19:21] Dependency 'puppetlabs-apt' (v1.6.0) would overwrite /etc/puppet/modules/apt [18:20:13] uses --target-dir and points to what puppet::self sets up [18:20:43] and that works [18:22:52] hmm, just not in /etc/puppet/modules/ Dependency 'puppetlabs-apt' (v1.6.0) would overwrite /etc/puppet/modules/apt [18:27:01] andrewbogott: How do you change your 'current project'? [18:27:50] The current project widget should be a popup... [18:27:56] Altnough maybe not if you're only a member of one project. [18:28:13] It isn't for me. Can you add me to tools? [18:29:38] Yep, done. [18:29:46] Although you're cloudadmin on wikitech-test, so you can do that yourself as well. [18:29:49] i installed the sensu module (monitoring solution pointed out by Ryan), but it wants now [18:29:52] This class requires puppet-apt module [18:30:26] now looks up puppet-apt itself .. hmm [18:32:21] andrewbogott: Oh, right, forgot that the wikitech thing was also there. :-) [18:32:44] andrewbogott: No popup; at least with chrome [18:32:52] * andrewbogott tries [18:33:15] * Coren attempts the burning canine. [18:34:51] No popup either. [18:36:14] puppetlabs actually uses HTTP status "410 - Gone" to tell me puppet/apt is not coming back, heh [18:37:43] Coren: http://bogott.net/misc/horiz.png [18:37:50] …is what it looks like for me in chrome [18:38:25] I don't know if you have to be projectadmin for it to appear [18:39:08] Yeah, I have the same general look but with no down arrow or popup. Lemme open the dev console to see if there is some javascript thingy. [18:41:52] 3Wikimedia Labs / 3deployment-prep (beta): Add RecentActivityFeed extension to beta labs - 10https://bugzilla.wikimedia.org/69785 (10Chris Steipp) [18:48:21] what exactly is expected when uploading my ssh key in the wikitech user preferences? is the output of cat the_key.pub correct? [18:48:59] hi does anyone know how to run phpunit tests on labs instances?... I tried physikerwelt@drmf:/vagrant/mediawiki$ sudo mwscript tests/phpunit/phpunit.php tests/phpunit/includes/parser/ [18:48:59] #!/usr/bin/env php [18:48:59] PHPUnit 4.1.4 by Sebastian Bergmann. [18:48:59] Class 'tests/phpunit/phpunit' could not be found in '/srv/vagrant/mediawiki/tests/phpunit/phpunit.php'. [18:48:59] somehow the required mwscript wrapper m [18:49:10] akes things comlicated [18:49:17] jongleur: yes [18:49:24] valhallasw`cloud: thanks [18:49:38] jongleur: the thing in the format ssh-rsa LONGSTRING some more text [18:49:38] (sorry for the linebreaks) [18:49:49] (or ssh-dsa, depending on the key) [18:50:00] valhallasw`cloud: done, thanks [18:50:59] physikerwelt: there seems to be a weird /srv in the error message [18:51:29] physikerwelt: your code is in /vagrant/mediawiki but the error says its looking in /srv/vagrant/mediawiki [18:51:46] physikerwelt: not sure *why*, though. Maybe try passing the full path to mwscript? [18:52:03] there is a symbolic link from /vagrant to /srv/vagrant [18:52:18] great, then I think I have to wait for the account approval [18:52:29] thanks for your help so far [18:52:48] physikerwelt: I see. [18:52:49] physikerwelt: hmm [18:52:54] valhallasw`cloud: the full path does not work either [18:53:16] mwscript automatically adds the maintenance path [18:53:17] physikerwelt: ln -s tests/phpunit/phpunit.php phpunit && mwscript phpunit [18:53:19] does that work? [18:53:32] maybe it's the class name that's wrong (the forward slashes do look weird) [18:54:39] valhallasw`cloud: mh..ln: failed to create symbolic link ‘phpunit’: File exists [18:55:20] physikerwelt: just the mwscript phpunit then :P [18:55:58] valhallasw`cloud: The MediaWiki script file "/vagrant/mediawiki/maintenance/phpunit" does not exist. [18:57:09] physikerwelt: err. You're not in the maintenance dir, are you? [18:57:19] or is the current phpunit a symlink there or something? [18:58:01] no I'm in the mediawiki root dir [19:01:12] valhallasw`cloud: is there a way to bypass the mwscript with a fake config [19:02:39] physikerwelt: I have no clue [19:02:46] physikerwelt: I'm just giving suggestions based on the error message [19:03:02] I haven't run MW tests in a year or two [19:04:27] valhallasw`cloud: OK. However, thanks for your help [19:08:38] * physikerwelt found a patch of bd808|LUNCH https://gerrit.wikimedia.org/r/#/c/154472/ [19:35:23] hi [19:36:06] is there an admin that can give "loginviashell" rights to Marcobra.ubuntu? [19:36:18] he just created an account [19:39:38] I'm the admin for https://wikitech.wikimedia.org/wiki/Nova_Resource:Osmit [19:41:40] simone_: Done. [19:41:56] scfc_de: thanks! [20:35:49] the guc tool is still not responding, can somebody perhaps restart the webservice? [21:16:38] !log sensu - project created - initial instance sensu-01 up [21:16:40] Logged the message, Master [21:17:00] !log sensu - installed sensu-puppet via puppet module install, checked puppet::self box [21:17:01] Logged the message, Master [21:17:32] !log sensu - complains puppet-apt is missing but at the same time i get that it is installed [21:17:33] Logged the message, Master [22:01:53] mutante, sensu is for where? [22:03:17] jeremyb: it's a monitoring framework [22:03:30] right [22:03:48] i.e. prod, labs, just to play, etc. [22:03:59] i hope there is no difference :p [22:04:12] well we already have separate nagios/ganglia/etc. [22:04:21] no, Ryan mentioned it, i just wanted to look at it [22:04:35] right. i think i heard of it before. and forgot [22:04:35] and also, _if_ we use something i hope it will be like it once was planned [22:04:39] test stuff in labs [22:04:43] then actually use it in prod [22:04:54] and have the same role in both realms.. imagine that:) [22:05:39] but that today was really just to look at it.. and also.. i just tried if it works to install stuff that way [22:05:44] puppet module install [22:05:46] i really hate that we need to get roots to merge stuff in order for it to be available for testing in labs. i guess we have puppetmaster::self. but not everyone could figure that out [22:06:00] i used puppet::self on that one [22:06:16] dont tell me to use vagrant please :) [22:06:31] i'm not! [22:06:36] you know we once had "test" branch [22:06:36] ? [22:08:03] jeremyb: the ::self stuff works mostly [22:09:13] mutante, yeah... i'm not advocating that either [22:09:33] mutante, we could e.g. have a puppet environment per openstack project [22:09:43] (just brainstorming, no idea if this is feasible) [22:09:58] I think we've gotten a pretty good workflow with puppet::self figured out for beta. I should really write up a nice wiki page and email about it [22:10:19] and then you could set your project's environment to point to a sandbox branch or whatever [22:10:29] or point to a real branch [22:11:09] We rebase on the production branch once an hour with a cron script to get new upstream changes. Local changes are made by submitting a patch to gerrit and cherry-picking to the puppet master [22:12:01] We occasionally get conflicts on the rebase that need to be manually resolved, but other than that it works smoothly [22:12:12] how do you find out if there are conflicts? [22:12:21] also, can has beta root again? [22:12:35] My cron script detects and rolls back [22:12:37] who else does this stuff [22:12:40] But it doesn't email yet [22:12:44] "running a [22:12:44] couple of sql scripts to add columns to meta's CentralNotice logging [22:12:45] tables" [22:12:48] I just check once a day or so [22:13:14] "a couple of scripts that change schema" ?:p [22:13:19] mutante, Reedy or springle ? are my best guesses [22:13:20] jeremyb: Anyone who has a signed NDA at the moment can have root in beta [22:13:23] ohh, columns [22:13:32] reedy does tables usually not columns [22:13:34] jeremyb: can it be true that "deployer" rights let you do this? [22:13:46] i doubt (or let's say hope not) [22:13:47] they let you create. not ALTER. i think [22:13:50] ah [22:14:13] where are you quoting from anyway? [22:14:17] mutante: The way I was told to do that was to submit a bug and assign to springle [22:14:39] Elliot wants permissions for that [22:14:40] https://wikitech.wikimedia.org/wiki/schema%20changes ! [22:14:48] and i wonder which admin group if any [22:15:08] yea, sounds like springle wants to now about that :) [22:15:24] * jeremyb hands mutante a k [22:15:39] i love those access requests that start with "need shell" [22:16:31] add column = schema change (to me) [22:16:34] or no [22:16:35] jeremyb: If you are interested, the script that we cron to rebase puppet is at -- https://github.com/wikimedia/operations-puppet/blob/production/modules/beta/files/git-sync-upstream [22:17:02] mutante: yes. that's a schema change [22:17:38] add, drop, modify, new index ... all schema changes [22:18:18] thanks for the wiki link, using that [22:18:32] legoktm: [22:18:38] :) [22:20:07] bd808, great [22:20:49] bd808: oh, I left a question for you on https://gerrit.wikimedia.org/r/#/c/155168/ about labs-vagrant [22:22:19] legoktm: Oh yeah. I pushed https://gerrit.wikimedia.org/r/#/c/155271/ to try and fix that [22:22:37] I forgot to put you on the review/ping you [22:22:40] just me or there's a much longer session expiration on gerrit vs. wikitech? [22:22:48] * legoktm tests [22:22:50] i would think those should be approximately in sync? [22:22:57] jeremyb: gerrit is 30 days I think, wikitech is 7 [22:23:09] gerrit seems longer, yea [22:23:31] I don't remember my Gerrit login ever expiring. Don't know about "drafts", though. [22:27:00] bd808: it's closer! [22:27:01] 'loadScript' => [22:27:01] string(41) "//gadgets-gadgets.wmflabs.org:/w/load.php" [22:27:07] the : probably shouldn't be there? [22:27:29] bah. That's because I changed the logic [22:27:47] it always expects a port now (didn't before) [22:27:54] THat needs thought [22:28:18] I'd like to keep it all in hiera but I may need to move that back out to puppet land [22:34:21] ok, I'll just hack around it for now [22:34:36] gadgets-gadgets.wmflabs.org and gadgetsclient-gadgets.wmflabs.org :/ [22:34:41] I probably could have picked shorter names... [22:38:30] i wonder if one day it will also be "php5-mariadb" for the PHP module to talk to db [22:38:54] since the other package names are adjusted [23:39:49] how to get https://forge.puppetlabs.com/puppetlabs/mysql#mysql_grant on labs? [23:50:52] !log wikistats - deleted old proxy, created new one, switched backend over to wikistats-petcow instance [23:50:53] Logged the message, Master [23:52:27] !log wikistats - so that now we have puppet status OK again (missing puppetized db setup) [23:52:27] Logged the message, Master