[00:19:20] milimetric, do you have a moment now? [00:19:30] sure thing jgonera [00:19:48] so you guys had an rsync moving the datafiles over after the SQL ran right? [00:19:57] so I tried doing full deploy using limn-deploy at http://mobile-reportcard.wmflabs.org/ [00:20:11] I don't know that, YuviPanda might [00:20:18] there was something in cron I guess [00:20:18] yeah, there's two parts [00:20:29] I'm pretty sure that cron synced the results of the SQL [00:20:42] I DIDN'T DO IT! [00:20:42] but if you update the graphs themselves, you'd have to redeploy, right [00:20:45] :) [00:20:46] it also had to run generate.py periodically [00:20:53] the data is sync'd by cron [00:20:56] that's in puppet [00:20:57] yep [00:21:16] so it sounds like limn-deploy is the part that's breaking now, let me check [00:21:30] not really, because the new graphs appeared [00:21:34] but they're empty now [00:21:40] and old graphs are not updated with new stats [00:21:47] milimetric, ^ [00:21:49] ah [00:21:50] I see [00:22:11] i'm gonna run this deploy to make sure [00:22:22] ok [00:22:32] i deployed to mobile-reportcard-dev, and it updated stuff [00:22:40] oh, but maybe you just skipped that? [00:22:47] (it's fine, that site's just for testing anyway) [00:23:09] I skipped it [00:23:21] k, I see that mobile-reportcard proper is up to date [00:23:29] on dev Successful contributions (per month) is also empty [00:23:43] and it also has a VE tab for some reason, but that's another weirdness [00:23:46] so it does look like the only problem could be the data synched by that cron [00:23:55] it seems so [00:24:54] be back soon [00:28:04] k, so the problem is that this 404s: http://stat1001.wikimedia.org/limn-public-data/mobile/datafiles/monthly-contributions.csv [00:28:17] but i can't figure out why the hell that VE tab is there [00:28:17] lol [00:28:18] That code was written by the last guy [00:34:33] ah, jgonera: figured out that mystery too [00:34:43] someone logged into limn0 and added commits to the repository there [00:35:06] so deploying does nothing because the merge probably always conflicts [00:36:15] but - the main problem is that the cron job broke somewhere along the way and didn't upload its newest files to stat1001 [00:36:35] YuviPanda: do you have time to look at that or you wanna show me where to poke? [00:36:48] the puppet cron? [00:36:49] I [00:36:53] didn't write that [00:36:57] I... think you did? [00:37:04] or someone from analytucs [00:37:38] also, if there are random uncommited things in the limn test repo related to VE, feel free to rm them - that was me [00:45:55] jgonera: I don't seem to have rights to check the status of that cron. I'll have to check with ottomata tomorrow. I've filed this bug and assigned it to Andrew so we don't forget: https://bugzilla.wikimedia.org/show_bug.cgi?id=56979 [01:30:30] milimetric, thank you [01:31:13] when limn-deploy stopped working, I manually logged into limn0 and just did git pull myself a couple of times, is this how I broke it? [17:39:25] lunch back in a bit [19:30:10] (PS1) JGonera: Fix error in edits_web table [analytics/limn-mobile-data] - https://gerrit.wikimedia.org/r/95202 [19:30:14] (PS1) JGonera: Remove performance graph [analytics/limn-mobile-data] - https://gerrit.wikimedia.org/r/95203 [19:30:28] (CR) JGonera: [C: 2 V: 2] Fix error in edits_web table [analytics/limn-mobile-data] - https://gerrit.wikimedia.org/r/95202 (owner: JGonera) [19:30:42] (CR) JGonera: [C: 2 V: 2] Remove performance graph [analytics/limn-mobile-data] - https://gerrit.wikimedia.org/r/95203 (owner: JGonera) [20:14:26] (PS6) Milimetric: Async cohort upload UI and fixes for validation [analytics/wikimetrics] - https://gerrit.wikimedia.org/r/93081 [20:15:27] that's what I love right there, feature implementation with net negative line counts [20:15:27] +586, -751 [20:15:29] :) [20:32:38] halfak: my bad, that field in logging is just a serialized php array, no JSON involved [20:33:04] Good to know. Now if I can find a library to help me parse it. :) [20:33:06] http://php.net/manual/en/function.serialize.php [20:33:31] a non-PHP library preferably. :\ [20:33:32] http://stackoverflow.com/questions/5935501/unserialize-php-data-in-python [20:33:34] ;) [20:33:38] Yup. [20:35:36] ottomata: could you elaborate on the relevance of openjdk to dclass? [22:11:53] oof jeremyb, barely [22:12:15] it had something to do with how the dclass package installed its .jars and where oracle jdk 6 wanted to load them from [22:21:41] ottomata: are we using dclass? will we be? [22:22:19] and only for analytics but not varnish? [22:22:34] e.g. we could do mobile redirect based on dclass [22:22:58] * jeremyb can't really talk atm though [22:25:35] (PS7) Milimetric: Async cohort upload UI and fixes for validation [analytics/wikimetrics] - https://gerrit.wikimedia.org/r/93081 [22:27:50] only for analytics [22:27:53] we could do with varnish [22:27:59] right now i only know of analytics use [23:13:11] (PS1) JGonera: Disable edits-monthly-5plus-editors graph updates [analytics/limn-mobile-data] - https://gerrit.wikimedia.org/r/95298 [23:13:41] jeremyb: I had a brief look at dclass once. Did you encounter problems with using dclass under openjdk? [23:13:42] (CR) JGonera: [C: 2 V: 2] Disable edits-monthly-5plus-editors graph updates [analytics/limn-mobile-data] - https://gerrit.wikimedia.org/r/95298 (owner: JGonera) [23:41:09] qchris: no. i never tried [23:41:42] jeremyb: Ah ok. [23:42:11] We currently use dclass to generate some files. [23:42:32] But I guess those files are currently not used by anyone. [23:42:57] And they currently are not in good shape (read: really bad shape)