[15:46:02] aight [17:32:21] milimetric: it is possible to get an update to http://mobile-reportcard.wmflabs.org/ ? [17:32:34] some changes haven't been deployed yet and i have no idea how to do that [17:33:08] grr [17:33:12] sorry jdlrobson [17:33:16] i'll update manually [17:33:20] milimetric: thanks [17:33:23] the deployer got hosed by some new labs update [17:33:27] (ssh issue I can't figure out) [17:33:35] so you have to ssh ino limn0 and pull the repo [17:33:50] i'll do so now [17:35:44] jdlrobson: if this looks good, I can pull the mobile-reportcard as well: http://mobile-reportcard-dev.wmflabs.org/ [17:35:54] * jdlrobson looks [17:36:23] looks good to me milimetric [17:36:56] jdlrobson: ok, pulled the real one [17:45:30] milimetric: thanks a bunch for the speedy deploy :D [17:45:43] np, sorry the automated one doesn't work [18:11:44] hey average, just a heads up, I'm removing all the unused imports [18:11:54] (checked it with snakefood and it's bothering me) [18:11:55] :) [18:29:52] snakefood ? [18:30:28] oh, just saw it in aptitude [18:39:51] drdee: I have a q about squid logs. are there some logs that are more detailed than what we have in squid logs ? [18:39:59] no [18:40:02] drdee: for example headers for each request are not present in squid logs [18:40:25] is this kind of data present in some form in a different place ? [18:41:03] average, you mean like, all http headers? [18:41:17] ottomata: or a subset [18:41:17] why do you need this if you are working on 704? [18:41:31] drdee: it's offtopic [18:41:35] well, i mean, the squid logs does have a subset of some headers [18:41:40] I mean, my question is offtopic [18:41:43] but no, we don't keep any more information than what is in those logs [18:41:56] ok [19:02:55] so drdee [19:03:01] yoooo [19:03:03] i'm back looking at this camus jar thing [19:03:10] oohohh [19:03:11] right now, it produces a shaded jar [19:03:16] with all the dependencies in it [19:03:20] i am slim shady! [19:03:21] is that the proper way to do things? [19:03:38] depends [19:03:47] if the dependencies are already installed on the system [19:03:52] they are not [19:03:55] plus, this is hadoop [19:03:59] then you can exclude them at compile time in the pom [19:04:16] oh it also makes non-shaded jars [19:04:20] none of the deps are installed on kraken? [19:04:24] yes it should create both [19:04:27] its just that the only one i've gotten to work are the shaded one [19:04:31] umm, i'm sure some are [19:04:38] yes that's what i would expect [19:04:40] like the avro stuff [19:04:45] but, kafka, gson, jodatime, etc. [19:04:46] avro [19:04:59] so you could exclude avor [19:05:12] by setting compile [19:05:14] i belief [19:05:23] qchris: is that correct ^^ [19:05:32] eh, one of the main classes we need has an explicit import for it [19:05:45] well, that's not true [19:05:57] the main class we need imports a linked in class that itself imports avro classes [19:06:53] (PS1) Milimetric: removing unused dependencies [analytics/wikimetrics] - https://gerrit.wikimedia.org/r/80431 [19:07:00] (CR) Milimetric: [C: 2 V: 2] removing unused dependencies [analytics/wikimetrics] - https://gerrit.wikimedia.org/r/80431 (owner: Milimetric) [19:07:06] but that should not matter ottomata [19:07:14] ok [19:07:18] its about where it needs to look for the dependency [19:07:22] as long as that class isn't instantiated, it won't matter? [19:07:33] and if the dependency is reachable on the the java path it should be fine [19:07:40] yes, i understand that [19:07:59] but do we want to try to pull all of these dependencies into hadoop? [19:08:01] or just use the shaded jar? [19:08:16] the shaded jar can cause dependency conflicts [19:08:30] if deps are both on the system and in the jar [19:08:48] so i would say embed only those deps that are not on kraken yet [19:09:33] hm [19:12:41] so [19:12:43] when I compile [19:12:56] i do get a lot of [WARNING] We have a duplicate javax/servlet/http/HttpServletRequestWrapper.class in .... [19:36:15] heya milimetric, yt? [19:44:41] milimetric: ^^ [19:50:15] hey Maryana, here [19:51:01] yo :) was just wondering why some of the graphs on http://mobile-reportcard.wmflabs.org/ were up to date, and some were cut off at august 13 [19:51:17] hm... [19:51:36] jdlrobson: do you know who updated the scripts / datafiles that back those graphs ^^ ? [19:52:06] I just push the code around, but I do know that the jobs run in prod somewhere and rsync the results [19:52:13] maybe they haven't rsync-ed or they failed... [19:52:23] i belief YuviPanda or juliusz [19:52:53] mostly juliusz of late, I've not touched those for a couple of months now [19:53:06] but... I did set them up initially (with help from milimetric :P) [19:54:04] ah Maryana, I think this is just Limn being confusing [19:54:14] August 13 is what it says in the Callout [19:54:37] but the data seems to be up through today on all the graphs [19:54:56] if you look at the X axis, you'll see it all goes past August 21st. [19:54:57] the one i care about specifically: http://mobile-reportcard.wmflabs.org/graphs/successful-edits-main [19:55:10] that one cuts off at tues 13 ^ [19:55:17] in the callout [19:55:25] um, no, I don't see that [19:55:28] i think dan is right [19:55:32] i see it go through to the 22nd [19:55:33] i will create a bug in bugzilla [19:55:36] me too [19:55:37] maybe do a hard refresh? [19:55:38] err, sorry, this one: http://mobile-reportcard.wmflabs.org/graphs/successful-edits-unique [19:55:43] but that's also cut off [19:55:49] yep, same thing [19:55:51] same ;) [19:55:53] ah! [19:55:58] i refreshed and it updated [19:55:59] hm... [19:56:04] it has a cache buster [19:56:04] sorry, false alarm! [19:56:10] so i'm confused why it hung on to the old data [19:56:12] no - it's still bad [19:56:24] I mean I definitely made it so it kills any cache... [19:56:56] https://bugzilla.wikimedia.org/show_bug.cgi?id=53232 [19:57:34] aha!! [19:57:40] the cache buster is broken for remote datafiles [19:57:45] :) Thanks Maryana, good catch! [19:58:13] * drdee awards Maryana the sharp eye award of the day [19:58:25] hm drdee, I think that bugzilla issue is not tractable [19:58:25] anytime :) [19:58:31] the callout has weird black magic in it [19:58:40] voodoo ? [19:58:51] but we should definitely add one for "cache-buster does not work on remote datafiles" [19:58:52] you promised me that Limn would no longer use voodoo! [19:59:00] dude, I tried :) [19:59:05] oh man! [19:59:29] you wanna update the bug report/ [19:59:30] ? [19:59:53] 53232 or you mean make one for the cache buster? [20:01:21] 53232, because that's what the end user sees [20:01:54] no, they're unrelated [20:02:02] i'll make a new one for the cache buster [20:02:13] and I'll edit the callout one [20:08:14] k [20:08:16] https://bugzilla.wikimedia.org/show_bug.cgi?id=53234 [20:08:21] drdee ^ [20:08:23] seen [20:16:47] milimetric: check https://mingle.corp.wikimedia.org/projects/analytics/cards/list?filters[]=%5BType%5D%5Bis%5D%5BDefect%5D&style=list&columns=Customer%2CDevelopment%20Status%2CEstimate%2CProject%20Name&tab=All [20:17:16] :) mingle seriously needs some prettier urls [20:17:17] geez [20:17:34] ah cool, so this is bingle? [20:17:39] i think it should be called bungle [20:18:48] yah bungle is also good [20:19:07] but that is bingle in action [20:19:49] hahaha [20:19:50] yeah bungle! [20:35:28] cool, drdee, maybe we should take out "Not Set" from Sprint planning? [20:35:34] and only toggle it when we run out of stuff [20:35:42] (the sprint planning view) [20:35:50] then that view would look a lot better [20:36:06] you sort of of need the not set column so you can easily move features in and out of a sprint [20:37:08] but shouldn't those features be roughly moved into sprints at the beginning, and just pushed back and forth in between sprints as we go? [20:37:31] and any new issues would just get added to the next or current sprint [20:38:16] we used to do that [20:38:22] but then as the release progresses [20:38:31] you have to move more and more cards to the next srpint [20:38:43] (unless we always deliver what we commit to0 [20:43:43] hey drdee [20:44:26] hey [20:44:46] sent you DM [20:44:50] replied :) [20:51:05] hey drdee [20:51:05] is your comscore account still active? [20:52:18] is it that time of the year again? [20:59:22] DarTar: my account has been disabled [20:59:24] oh wait, it's alright – I got a strange message like "your account has been disabled", but I managed to get in again [20:59:24] not sure why [20:59:27] ha [20:59:44] try and insist a couple of times with the Forgot Password form [20:59:50] eventually it'll give up [21:00:05] I reset mine, and it's working fine now [21:00:10] :D [21:01:41] mine is still disabled [21:01:46] time for a phone call :) [21:12:38] drdee: I was looking up the latest UV data (July 2013), Yahoo had a tremendous YoY growth (+20%) in the US while Wikimedia properties remained virtually flat in the US, but worldwide Wikimedia has the strongest growth (+8.9%) of all top properties: Google is at +3%, Yahoo +6.6% and FB -6.2% [21:13:31] FB: -13.5% in the US as of July - w00t [22:49:00] average, btw, your way of running nosetests saves me so much time :) [22:49:10] now if we cache the stupid project host map we'll be flying [23:29:53] (PS1) Milimetric: even more tests [analytics/wikimetrics] - https://gerrit.wikimedia.org/r/80516 [23:30:03] (CR) Milimetric: [C: 2 V: 2] even more tests [analytics/wikimetrics] - https://gerrit.wikimedia.org/r/80516 (owner: Milimetric) [23:30:12] :) broke 100 tests