[04:55:24] (PS15) Mforns: Add ability to global query a user's wikis [analytics/wikimetrics] - https://gerrit.wikimedia.org/r/129858 (owner: Terrrydactyl) [07:21:23] Analytics / Refinery: Make webrequest partition validation handle races between time and sequence numbers - https://bugzilla.wikimedia.org/69615#c15 (christian) Happened again for: 2014-10-24T19/2H (on bits) [07:23:27] !log Marked raw bits webrequest partitions for 2014-10-24T19/2H ok {{bug|69615#c15}} [07:53:25] Analytics / Refinery: Raw webrequest partitions that were not marked successful - https://bugzilla.wikimedia.org/70085 (christian) [07:53:27] Analytics / Refinery: Raw webrequest bits partition for 2014-10-26T21/1H not marked successful - https://bugzilla.wikimedia.org/72548 (christian) NEW p:Unprio s:normal a:None The bits webrequest partition [1] for 2014-10-26T21/1H has not been marked successful. What happened? [1] _______... [08:22:23] Analytics / Refinery: Raw webrequest bits partition for 2014-10-26T21/1H not marked successful - https://bugzilla.wikimedia.org/72548#c1 (christian) Only cp3019 is affected. For that host data worth ~55 seconds got lost in the ~1 minute between 2014-10-26T21:16:22 2014-10-26T21:17:24. I could neither... [08:31:53] Analytics / Refinery: Raw webrequest bits partition for 2014-10-26T21/1H not marked successful - https://bugzilla.wikimedia.org/72548#c2 (christian) Kafka logs did not show peculiar entries in the relevant period of time. [08:39:23] Analytics / General/Unknown: Kafka broker analytics1021 not receiving messages every now and then - https://bugzilla.wikimedia.org/69667 (christian) [08:39:26] Analytics / Refinery: analytics1021 getting kiked out of kafka partition leader role on 2014-10-27 ~07:12 - https://bugzilla.wikimedia.org/72550 (christian) NEW p:Unprio s:normal a:None analytics1021 again got kicked out of it's kafka partition leader role on 2014-10-27 ~07:12. I am not ru... [10:32:52] Analytics / Tech community metrics: Automate creating charts from Bugzilla Weekly Report - https://bugzilla.wikimedia.org/49744#c6 (Željko Filipin) Should this bug be then resolved as wontfix? [13:47:47] (CR) Ottomata: [C: 2] Ignore Maven output [analytics/refinery/source] - https://gerrit.wikimedia.org/r/168775 (owner: QChris) [13:47:52] (CR) Ottomata: [V: 2] Ignore Maven output [analytics/refinery/source] - https://gerrit.wikimedia.org/r/168775 (owner: QChris) [13:48:40] (CR) Ottomata: [C: 2 V: 2] Ignore Eclipse settings [analytics/refinery/source] - https://gerrit.wikimedia.org/r/168776 (owner: QChris) [13:57:17] apparently linux is shy - wifi refuses to connect anywhere but my house [14:02:08] ottomata: sneeeeeaaaaky meeting time :-D [14:02:36] ack! [14:14:53] milimetric: I'd like to deploy to staging too, can we do it together? [14:15:00] yes, def. [14:15:08] i'll ping you when i finish the review [14:18:01] mforns: for https://gerrit.wikimedia.org/r/#/c/129858/14..15/wikimetrics/controllers/cohorts.py,cm my comment from patchset 14 is wrong right? [14:18:15] as in, you need to set the centralauth_service there so you can use it in the cohort_upload, right? [14:19:28] I suppose the line is needed, am I wrong? [14:19:52] milimetric: ^ [14:20:05] mforns: no, my comment is wrong [14:20:49] feel free to reply saying that in the future - i'm wrong a lot :) [14:22:54] mforns: sorry I realized I forgot something [14:23:08] you'd have no way of knowing this from the code, but there are two more changes needed for a new session [14:23:13] xD [14:23:16] controllers/request_interceptors.py [14:23:32] that cleans up sessions on request teardown [14:24:14] ok [14:24:18] I'll have a look [14:24:36] mforns: the other one is task_postrun in configurables.py [14:24:44] does the same thing, but in the context of a celery task [14:25:21] ok [14:25:22] hm, those should really be calling the same function - ew (but don't change that, we should fix it separately) [14:26:26] fine [14:31:26] (CR) Nuria: [C: 2] Fix test and empty selection cases [analytics/dashiki] - https://gerrit.wikimedia.org/r/168414 (owner: Milimetric) [14:31:34] (CR) Nuria: [V: 2] Fix test and empty selection cases [analytics/dashiki] - https://gerrit.wikimedia.org/r/168414 (owner: Milimetric) [14:40:04] (PS16) Mforns: Add ability to global query a user's wikis [analytics/wikimetrics] - https://gerrit.wikimedia.org/r/129858 (owner: Terrrydactyl) [14:40:23] milimetric: pushed it [14:44:19] milimetric, nuria: Is it OK for you if I go for lunch now? We must go to the restaurant, cause we have no kitchen stuff any more :( [14:44:42] of course mforns let's push when you're back [14:44:47] no hurry, i have a million things [14:44:50] and everything looks great [14:45:08] ok, so see you in 1 hour [15:15:45] heading to a cafe [15:53:36] qchris: where is that hdfs fuse mount thing? [15:53:48] milimetric: /mnt/hdfs on stat1002 [15:53:51] thnx [15:53:54] yw [15:55:54] qchris: and is that where the projectcount files are? If so, where? Feel free to point me at documentation if that exists [15:56:44] milimetric: /mnt/hdfs/wmf/data/archive/webstats [15:56:53] Docs are at https://wikitech.wikimedia.org/w/index.php?title=Analytics/Pagecounts-all-sites#stat1002.eqiad.wmnet [15:57:19] But feel free to ping me :-) [15:59:41] Analytics / Wikimetrics: Tasks should clean up after themselves when SoftTimeLimitExceeded exception is raised - https://bugzilla.wikimedia.org/72563 (nuria) NEW p:Unprio s:normal a:None Tasks should clean up after themselves when SoftTimeLimitExceeded exception is raised. If not a report... [16:04:28] (PS1) Yurik: log2dfs script [analytics/zero-sms] - https://gerrit.wikimedia.org/r/168980 [16:05:07] Analytics / Wikimetrics: Tasks should clean up after themselves when they fail in STARTING phase - https://bugzilla.wikimedia.org/72563 (nuria) [16:06:22] Analytics / Wikimetrics: Tasks should clean up after themselves when they fail in STARTING phase - https://bugzilla.wikimedia.org/72563#c1 (nuria) >Tasks should clean up after themselves when SoftTimeLimitExceeded exception is raised. Taking this back, this is already happening. Now, what is missing is... [16:08:23] Analytics / Wikimetrics: Tasks should clean up after themselves when they fail in STARTING phase - https://bugzilla.wikimedia.org/72563#c2 (nuria) We likely need to catch all db errors via catching the highest level exception on sqlalchemy exception classes at the same level where we are catching now... [16:11:27] (PS2) Yurik: log2dfs script [analytics/zero-sms] - https://gerrit.wikimedia.org/r/168980 [16:11:47] (CR) Yurik: [C: 2 V: 2] log2dfs script [analytics/zero-sms] - https://gerrit.wikimedia.org/r/168980 (owner: Yurik) [16:30:22] (PS1) Yurik: log2dfs fix pattern match picker [analytics/zero-sms] - https://gerrit.wikimedia.org/r/168986 [16:30:22] nuria__: disabling puppet on staging disables the agent entirely [16:30:36] so we'd have to run puppet --enable, puppet agent -tv, puppet --disable [16:30:45] is that what you'd like to do? I'm not a fan personally [16:31:24] (CR) Yurik: [C: 2 V: 2] log2dfs fix pattern match picker [analytics/zero-sms] - https://gerrit.wikimedia.org/r/168986 (owner: Yurik) [16:40:29] ottomata, what's the control on when old Hive data gets scrubbed? [16:40:36] like, is there a task once a day, is it every 30 days... [16:40:56] runs every 4 hours [16:41:00] no [16:41:01] sorry [16:41:03] ah yes [16:41:04] every 4 hours [16:41:12] gotcha [16:41:15] deletes all partitions older than 31*24 hours [16:41:42] https://github.com/wikimedia/operations-puppet/blob/production/manifests/role/analytics/refinery.pp#L87 [16:45:35] aha, 31! Ta [17:02:57] milimetric: if you do not disable puppet, it will restart queue everytime it runs, (why?) [17:03:03] killing all running jobs [17:04:27] sounds like a puppet/upstart problem [17:04:42] maybe puppet doesn't know how to properly detect that queue is running [17:08:12] ottomata: it's upstart the one doing the detection, right? [17:08:52] well, puppet will attempt to start it if it thinks it is down [17:08:52] OR [17:09:13] there are some params to the service resource that are sometimes relevant [17:09:14] e.g. [17:09:16] hasstatus [17:09:21] but ja [17:09:22] OR [17:09:28] it could be the service is subscribed to a config file [17:09:32] that is changing on every puppet run [17:09:44] this happens, say, if an array or hash is iterated through without sorting first [17:09:50] well, hash, not array [17:10:01] because has keys are not guarunteed to be sorted or iterated thorugh the same way each time [17:15:06] ottomata: one question, in operations/puppet/manifests/role/wikimetrics.pp says that "This class does not (yet) work in production!" and "wikimetrics does not yet run via puppet in production" [17:15:53] I'm trying to add centralauth user, pass, host and name to production repo so that wikimetrics puppet module can use it [17:15:56] yes, production != labs [17:15:58] in that context [17:16:12] ? [17:16:22] when you say production repo, what do you mean? [17:16:31] operations/puppet [17:16:51] not mediawiki/vagrant/puppet [17:17:13] ah [17:17:16] k, puppet repo :) [17:17:21] hehe ok [17:17:21] yes, don't worry about that production block [17:17:32] that is just to keep someone from applying the wikimetrics role on a bare metal node [17:17:39] rather than in the virtual labs environment [17:17:55] 'wikimetrics production' is a different concept than 'ops production' [17:18:03] you want the stuff in the 'labs' blok [17:18:06] block* [17:18:10] ok, milimetric and I just added the fields to the file in staging and it worked [17:18:21] i'm not sure what value for centralauth db to put there [17:18:28] aha [17:18:41] for name and host [17:18:44] yeah, no idea [17:18:52] you'll have to ask milimetric where that info comes from i think [17:19:07] there is a passworrds.pp file that is rebased in staging and production that I thikn we store tthose things in [17:19:10] its manually on the labs instances [17:19:13] (he knows, we just manually configured staging [17:19:14] not pushed to any remote [17:19:14] ) [17:19:15] ok cool [17:19:19] we hardcoded, s1.labsdb/centralauth_p it for testing and it worked. [17:19:31] milimetric: you know about the rebase thing, right? [17:19:54] i have to get lunch before ops meeting, i ahve to run [17:19:56] will be back later [17:19:57] sorry guys [17:20:05] when we pull the puppet repo, we do it with rebase, so it doesn't bury our important passwords commit [17:20:05] yep [17:20:15] bye [17:20:49] ok, pull --rebase [17:21:39] Analytics / Wikimetrics: Puppet is restarting queue everytime it runs - https://bugzilla.wikimedia.org/72565 (nuria) NEW p:Unprio s:normal a:None Puppet is restarting queue everytime it runs on staging. Queue is running so it should not need a restart. [17:21:41] milimetric: so I just let the values hardcoded in the pp file? [17:21:52] Analytics / Wikimetrics: Puppet is restarting queue every time it runs - https://bugzilla.wikimedia.org/72565 (nuria) [17:22:02] ottomata: filed bug with upstart problem but - as far as i can rememeber- [17:22:08] that was not happening before [17:22:10] he's gone [17:22:12] ottomata: https://bugzilla.wikimedia.org/show_bug.cgi?id=72565 [17:22:18] will be back after ops meeting [17:22:50] rightttt [17:38:14] milimetric: I'm going to set also a db_pool size for centralauth, do you think stag=>10 and prod=>20 is ok? [17:38:42] didn't you set the pool to NullPool? [17:39:31] yeah, mforns: https://gerrit.wikimedia.org/r/#/c/129858/16/wikimetrics/database.py,cm [17:39:47] setting the pool class to NullPool means you're not using a pool to connect to CA [17:40:25] which makes sense. We'll replace all the labsdb connections with a single unified pool of connections in a future change [17:41:31] yes... [18:11:24] (PS17) Mforns: Add ability to global query a user's wikis [analytics/wikimetrics] - https://gerrit.wikimedia.org/r/129858 (owner: Terrrydactyl) [18:34:25] * milimetric turns on hesitant mode [18:34:44] qchris: is it possible you guys forgot to de-duplicate en, En, and EN in the projectcounts? [18:34:53] :-) [18:35:00] No, we did not forget. [18:35:03] i read the wiki page but couldn't find info [18:35:18] It's one of the idiosyncrasies of webstatscollector. [18:35:30] We wanted to reproduce that in the Hive files, so [18:35:47] people can really compare apples to apples when comparing hive and C implementation of webstatscollector. [18:36:01] gotcha, ok, but there's not some secret EN.wikipedia site, it's all the same :) [18:36:15] It's all the same :-) [18:36:33] thanks, I'll add a note to the pagecounts-all-sites article [18:36:42] Awesome! [19:07:31] DarTar: yt? [19:07:58] ottomata: hey [19:08:17] hey, just bumping the 'who should be in the researcher group' question [19:08:21] i'm not really sure of the current status [19:08:25] but i'm on RT duty this week [19:08:28] so i'd really like to resolve that [19:09:01] ottomata: I saw that thread, sorry for not responding earlier. I was also hoping for a final answer to the question of individual SQL credentials [19:09:08] which might make the whole proposal obsolete [19:09:23] I’d be strongly in favor of that option [19:09:29] but it was first introduced by ops [19:09:40] it sounds like it isn't going to happen [19:09:43] at least not in the near term [19:10:28] alright, I need to jump into an urgent check-in with halfak and leila, I’ll come back in a moment [19:10:59] ok [19:32:05] Could someone please review this? https://gerrit.wikimedia.org/r/#/c/168202/ thanks [19:45:53] ottomata: I’ll follow up on the thread re: SQL creds [19:46:06] ok [19:46:07] thanks [19:53:14] (CR) Ottomata: Add UAParserUDF from kraken (1 comment) [analytics/refinery/source] - https://gerrit.wikimedia.org/r/166142 (owner: Ottomata) [19:57:06] bmansurov: looking [19:58:12] (PS2) Nuria: Show namespace field placeholder next to the field [analytics/wikimetrics] - https://gerrit.wikimedia.org/r/168202 (https://bugzilla.wikimedia.org/71582) (owner: Bmansurov) [20:00:19] thanks nuria__ [20:14:07] (CR) Nuria: Show namespace field placeholder next to the field (1 comment) [analytics/wikimetrics] - https://gerrit.wikimedia.org/r/168202 (https://bugzilla.wikimedia.org/71582) (owner: Bmansurov) [20:15:37] (PS1) Ottomata: Bump version to 1.3.0-wmf1 [analytics/ua-parser] (wmf) - https://gerrit.wikimedia.org/r/169204 [20:15:44] qchris: ^ [20:16:30] * qchris clones and looks [20:21:49] (CR) QChris: "Not sure since upstream is in 1.3.0-SNAPSHOT although 1.3.0" [analytics/ua-parser] (wmf) - https://gerrit.wikimedia.org/r/169204 (owner: Ottomata) [20:22:10] hm, maybe so, although there isn't a 1.3.1 upsream, qchris [20:22:24] i could make a 1.3.1wmf-wmf1? [20:22:26] tag? [20:22:27] or [20:22:30] a 1.3.1wmf tag [20:22:36] dunno [20:22:37] as I said before ... the java part of ua-parser is nasty. [20:22:55] i think since there isn't an upstream version increase [20:23:05] I am fine with 1.3.0-wmf1, if we want to. [20:23:08] we can just make wmf1 be our release of the current master? [20:23:13] k [20:23:19] (CR) QChris: [C: 2 V: 2] Bump version to 1.3.0-wmf1 [analytics/ua-parser] (wmf) - https://gerrit.wikimedia.org/r/169204 (owner: Ottomata) [20:23:32] danke! [20:25:29] qchris, so patch it! [20:25:34] I'm happy to +2 the patches ;p [20:25:43] Naw. You abandoned mine :-P [20:25:57] (PS11) Nuria: Add UAParserUDF from kraken [analytics/refinery/source] - https://gerrit.wikimedia.org/r/166142 (owner: Ottomata) [20:25:58] Or ... the one's I cared about :-) [20:26:27] Meh. I don't care actually. [20:27:50] (CR) Nuria: Add UAParserUDF from kraken (1 comment) [analytics/refinery/source] - https://gerrit.wikimedia.org/r/166142 (owner: Ottomata) [20:41:27] nuria__: was talking with qchris about ua parser stuff [20:41:37] he wants to have some process around keeping the ua-parser dependency up to date [20:41:39] ottomata: aham [20:41:52] i told him that's a good idea, but we shoudln't make the process definition block this change [20:41:53] ottomata: besides the parametric test? [20:41:56] yes [20:42:08] but, we compromised by saying I would update to latest ua-parser regexes [20:42:14] and add a new jar to our archiva [20:42:19] and build our udf against that [20:42:22] it works, but now tests fail. [20:42:27] ottomata: aha , testing it would be a matter of updating the pom [20:42:36] do you mind if I commit the pom change, and you fix the tests? [20:42:48] ottomata: no, that's fine, it's expected [20:42:52] ok [20:43:51] ottomata: ua parser classifications are expected to change as they correct/update data all the time [20:44:10] aye [20:44:10] (PS12) Ottomata: Add UAParserUDF from kraken [analytics/refinery/source] - https://gerrit.wikimedia.org/r/166142 [20:44:19] ok cool, there you go. grab that and try to build [20:46:14] ottomata: k [20:52:56] tnegrin: I don't understand how the star/snowflake schema would help with our data spew... unless you're suggesting that we store the raw data without aggregating it? [20:54:54] The problem I see is, we have thousands of impressions per minute, and we have to aggregate along various dimensions. So we either have to store raw requests, or perform all the possible aggregations and store that in that form. [21:04:50] ottomata, qchris: so new ua parser is reporting ""Mozilla/5.0 (Windows NT 6.3) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/38.0.2125.104 Safari/537.36"" [21:05:06] as being "Windows 7" [21:05:43] qchris, ottomata; ahem ...isn't that wrong? [21:06:03] It should be Windows 8.1 ... shouldn't it? [21:06:16] i saw 8.1 in the tests [21:06:34] mostly [21:06:34] OS name check expected: but was: [21:07:01] Argh, sorry yes, [21:07:08] ottomata is correct [21:08:19] ottomatta, but little did we know it ua parser is right [21:08:46] this is ie11 in windows 8 [21:08:47] Mozilla/5.0 (Windows NT 6.3; Trident/7.0; rv:11.0) like Gecko [21:11:25] Analytics / Wikimetrics: Windows Line endings are not handled properly - https://bugzilla.wikimedia.org/72581 (Dan Andreescu) NEW p:Unprio s:normal a:None This was a problem before, but the fix was not comprehensive. On input from windows machines, wikimetrics ends up creating blank lines... [21:13:56] Analytics / Wikimetrics: User validation throws exception when accessing an unknown project - https://bugzilla.wikimedia.org/72582 (Dan Andreescu) NEW p:Unprio s:normal a:None If a project that isn't replicated to labs is passed in as a user's project (example: wikimania2014wiki), the valid... [21:28:52] (CR) Bmansurov: Show namespace field placeholder next to the field (1 comment) [analytics/wikimetrics] - https://gerrit.wikimedia.org/r/168202 (https://bugzilla.wikimedia.org/71582) (owner: Bmansurov) [21:30:10] Analytics / Wikimetrics: Windows Line endings are not handled properly - https://bugzilla.wikimedia.org/72581 (Marcel Ruiz Forns) NEW>ASSI a:Marcel Ruiz Forns [21:35:32] (PS1) Mforns: Normalize windows line endings in cohort csvs [analytics/wikimetrics] - https://gerrit.wikimedia.org/r/169223 (https://bugzilla.wikimedia.org/72581) [22:16:45] (CR) Nuria: "If you look at the change (after I rebased it) you will see how the text appears twice. Should be removed from text input if that was the " [analytics/wikimetrics] - https://gerrit.wikimedia.org/r/168202 (https://bugzilla.wikimedia.org/71582) (owner: Bmansurov) [22:21:31] (PS3) Bmansurov: Show namespace field description next to the field and remove it from placeholder [analytics/wikimetrics] - https://gerrit.wikimedia.org/r/168202 (https://bugzilla.wikimedia.org/71582) [23:57:13] (PS13) Nuria: Add UAParserUDF from kraken [analytics/refinery/source] - https://gerrit.wikimedia.org/r/166142 (owner: Ottomata)