[00:00:37] madhuvishy: I worked on the other one I am doing with Marcel [00:00:58] I don't know how to test on beta cluster though, because it doesn't seem to have zookeeper [00:01:07] milimetric: alright. let's catch up tomorrow and pair on our thing. [00:01:08] aah [00:01:19] milimetric: I'm trying to test the oozie thing i wrote [00:01:23] sure, let's do the config one tomorrow [00:01:25] feel better [00:01:32] milimetric: its blah [00:02:16] milimetric: oh also, is there a way to get staging wikimetrics database on local? [00:03:41] madhuvishy: what do you mean? [00:03:47] download all the data? [00:03:56] milimetric: yeah [00:04:08] why do you need that? [00:04:35] sorry, yes, there's a way, but I'm just making sure there's not an easier way to do what you want [00:04:55] milimetric: i thought it might be nice to test the validate again change on that [00:04:59] to dump a mysql db, you can just do mysqldump [options] -Dwikimetrics > wikimetrics.backup.sql [00:05:13] oh, madhuvishy, yeah, then just deploy it to staging [00:05:20] no need to copy the db locally [00:05:31] it wouldn't help you anyway, because you don't have access to labsdb [00:05:35] milimetric: alright. [00:05:48] how do i deploy to staging - is there a wiki on that [00:06:07] there... might be? But it's easy [00:06:15] wikimetrics is deployed to /srv/wikimetrics [00:06:25] so just go there, sudo git checkout your gerrit change [00:06:39] milimetric: okay [00:06:43] will do [00:06:47] run whatever you need to run (alembic upgrade, install script, etc.) [00:06:47] thanks :) [00:06:51] hang on, not done :) [00:06:55] milimetric: aaah [00:06:56] :D [00:06:58] okay [00:07:23] then you restart the services like wikimetrics-queue, wikimetrics-scheduler, and apache. [00:08:04] if you need to change the configuration in /etc/wikimetrics then let me know I have to show you how to disable puppet (because it will overwrite the config every few minutes when it's running) [00:08:49] and that should be it, you can tail the logs in /var/log/apache2 and /var/log/upstart/wikimetrics-queue [00:08:49] milimetric: no config changes necessary i think - in this case - but useful to know that. [00:08:59] milimetric: Awesome. [00:09:07] k, let me know if you run into trouble [00:09:14] oh, madhuvishy [00:09:21] one thing when you rebase [00:09:23] there's a hidden conflict [00:09:35] because Marcel's code that got merged while yours was in review has an alembic migration [00:09:48] so you'll have to update your migration to have his as the "down" revision [00:09:52] milimetric: aah [00:10:19] i recommend rebasing before deploying to staging [00:10:28] milimetric: yeah i'll do that [00:10:41] but then if something breaks and it ends up being Marcel's code, just yell at me for it 'cause I merged it [00:11:33] ha ha okay [00:14:04] milimetric, do you know if External Store is available on stat1003? [00:14:41] matt_flaschen: i don't know what External Store is [00:14:46] oh, where rev text is? [00:14:50] milimetric, got attacked by a yuvipanda [00:14:58] omg halfak are you ok? [00:15:05] I'll survive. :) [00:15:09] phew [00:15:18] milimetric, yeah. https://wikitech.wikimedia.org/wiki/External_storage [00:15:40] So, we should probably delay a conversation about VE work until tomorrow given that (a) it's late in EDT and (b) I have to prepare to give a presentation tomorrow. [00:15:49] matt_flaschen: that's like a black box to me, I don't know if it's replicated [00:16:07] milimetric, okay, no problem. [00:16:10] matt_flaschen: thanks for pointing me to the article, I never knew what this thing was called [00:16:30] halfak: cool, I've got a little vega presentation I'm trying to help with too. let's talk tomorrow [00:16:33] I just like to use research databases for doing reads in production, just in case. I'm assuming these are read-only... [00:16:39] milimetric, I found a slot on our calendars. [00:16:46] 5:30 EDT [00:16:47] OK? [00:16:51] halfak: works [00:16:57] matt_flaschen: read only, yes [00:16:58] Great. Thanks [00:17:02] halfak, do you ever do any research with the DB that involves page text? [00:17:11] Because that must involve these tables [00:18:47] matt_flaschen: yes he does, he gets the text from dumps though [00:18:57] Yeah, so not direct DB access. [00:18:58] we were told External Store was a beast to access [00:19:11] It may not be replicated. [00:19:19] except it was more like "That ^&* &#*#O* no way man &**((&" [00:19:27] milimetric, LOL. [00:19:56] Not sure why, other than the cluster name changing every so often, it should still be replicatable. Though there are some other complexities with decoding it (gzip, etc.). [00:24:26] from what I understand if you want a specific revision and it's far back in the past, it's stored together with other revisions in blobs and there's magic code that knows how to get it out, but I was told that code was pretty hard to port / understand [00:24:40] I doubted this, because I can take some guesses as to who wrote that code and they're all fine engineers [00:24:51] but I never looked at it [01:11:56] (PS1) Madhuvishy: Add generic oozie component for emailing data [analytics/refinery] - https://gerrit.wikimedia.org/r/210632 [01:15:40] (PS2) Madhuvishy: Add generic oozie component for emailing data [analytics/refinery] - https://gerrit.wikimedia.org/r/210632 [08:47:59] Analytics, Engineering-Community, ECT-May-2015: Analytics Team Offsite - Before Wikimania - https://phabricator.wikimedia.org/T90602#1281713 (Qgil) [09:46:30] (CR) Mforns: Permit rerun of failed report (1 comment) [analytics/wikimetrics] - https://gerrit.wikimedia.org/r/208017 (https://phabricator.wikimedia.org/T88610) (owner: Mforns) [09:49:46] (CR) Mforns: Add error message to failed reports (1 comment) [analytics/wikimetrics] - https://gerrit.wikimedia.org/r/203241 (https://phabricator.wikimedia.org/T88610) (owner: Mforns) [11:13:03] milimetric: Hi [11:13:09] Let me know when you're around [11:13:14] hi joal|night [11:13:24] AH, I am still nighty [11:13:27] Let's change that [11:13:27] :) [11:13:44] * joal feels awake now [11:13:50] heh [11:13:58] I have a zookeeper address for you :) [11:14:01] yay! [11:14:10] And a few hints as well :) [11:14:18] yay^2 [11:14:22] huhu [11:14:48] I wonder, does yay^2 == yayyay ? [11:14:58] :) [11:15:18] So, kafka in betalabs [11:15:28] --> deployment-zookeeper01:2181/kafka/deployment-kafka [11:15:43] sorry, zookerper in betalabd, for jafja [11:16:17] I'm assuming jafja is kafka [11:16:36] Yes, you are right :) [11:16:44] I like jafja as well :) [11:16:45] what's the /kafka/deployment-kafka thing? [11:16:54] jafja should totally be a thing, we should build it [11:17:11] It's the zookeeper inner path for kafka values [11:17:40] ah, I see, so I pass this to --zookeeper deployment-zookeeper01:2181/kafka/deployment-kafka [11:17:42] like that right? [11:17:48] correct [11:17:50] k, adding to wikitech [11:18:09] hmmm ! didin't know there was a wikitech page on this ! [11:18:11] https://wikitech.wikimedia.org/wiki/EventLogging/Testing/BetaLabs#Kafka [11:18:17] Nice ! [11:18:32] well, I didn't know the root password for the beta cluster db was there [11:18:38] Another way to have the same effect : export ZOOKEEPER_URL=deployment-zookeeper01:2181/kafka/deployment-kafka [11:18:48] btw, I was told it's "Beta Cluster" not "Beta Labs" because it's managed by different people that aren't labs ops [11:18:54] (release eng) [11:19:49] I think we should rephrase the kafka stuff here [11:20:20] When using kafka on betalabs: deployment-kafka02.eqiad.wmflabs [11:20:30] beta cluster :) [11:20:39] Oops, sorry [11:20:43] buffer not updated yet [11:20:46] :) [11:20:47] heh [11:21:07] When using kafka on betalabs: deployment-kafka02.eqiad.wmflabs, using zookeeper deployment-zookeeper01:2181/kafka/deployment-kafka [11:21:28] I'm gonna take out the trash, feed the animals, and do other morning things. I'll be back to try out my new zookeeper, thanks for that [11:21:38] no prob [11:22:11] Let me know when you come back, I will show you where we are with Marcel (if you wish [14:06:26] joal, I'm a little confused about the data flow on beta [14:06:37] hm [14:06:41] i see a trickle of server side events [14:06:42] batcave ? [14:06:45] sure [15:01:42] o/ joal [15:01:52] Hi halfak [15:02:08] Now a good time to chat? [15:02:13] I am in a talk with Dan :) [15:02:26] OK. I have the next hour blocked off to sync up. [15:02:34] Could you ping me when you are ready? [15:02:57] Ready :) [15:03:07] halfak: --^ [15:03:19] Where do we meet ? [15:03:55] batcave? [15:04:29] Sure [15:04:31] OMW [15:19:58] milimetric: fighting git [15:24:12] joal: I can help with that [15:24:14] to the batcave! [15:24:19] OMWn [15:43:48] Analytics-EventLogging, Analytics-Kanban: ContentTranslationError event logging table is not receiving new events - https://phabricator.wikimedia.org/T98842#1282359 (ggellerman) a:Milimetric [15:47:37] Analytics-EventLogging, Analytics-Kanban, Patch-For-Review: Troubleshoot EL performance problems on 2015-05-06 - https://phabricator.wikimedia.org/T98588#1282384 (mforns) [15:47:38] Analytics-EventLogging, Analytics-Kanban: ContentTranslationError event logging table is not receiving new events - https://phabricator.wikimedia.org/T98842#1282383 (mforns) [16:50:43] Analytics-Kanban: Deal with non-timeboxed queries recomputing too much data - https://phabricator.wikimedia.org/T98979#1282555 (Milimetric) NEW a:Milimetric [16:51:11] (PS2) Milimetric: Disable 19 queries from the scheduler [analytics/limn-mobile-data] - https://gerrit.wikimedia.org/r/210364 (https://phabricator.wikimedia.org/T98979) [16:59:03] hello milimetric. do you know if we will be having the Analytics checkpoint tomorrow? :-) [16:59:29] I'd like to attend it, but if it's not happening, I need to schedule another meeting at that time. [16:59:59] lzia: joal says he can be there and hang out with you guys, but maybe it's up to you [17:00:01] Andrew won't be around [17:01:44] it's good to hangout with joal. okay, I'll plan to attend the first half hour. thanks milimetric. [17:12:41] milimetric, hi, did you talk to veag? [17:12:42] vega [17:12:58] milimetric, want to present something yourself tomorrow? [17:13:06] hehe, two q at once [17:13:42] yes, and maybe? I've been looking at lyra [17:14:23] yurik: http://idl.cs.washington.edu/projects/lyra/ [17:14:37] I wanted to basically walk people through looking at the same data in a few different ways [17:14:51] and try to make it as smooth as possible. And as usual, I got stuck on some Vega syntax [17:15:11] so I was thinking this tool might be useful as it generates vega definitions [17:15:35] it's pretty well thought out, but still confusing :) [17:15:46] WOWOOWOWOW [17:15:52] so basically, right now, I'd have nothing great to show but I can show this [17:16:04] visual :) [17:16:09] it's nice, yea [17:16:17] nah, don't, because its not generating vegas yet, right? [17:16:23] it is generating vega [17:16:26] you just go "export" [17:16:40] you can also import, of course [17:16:47] it does? [17:16:49] holy crap [17:16:51] yes [17:16:52] :) [17:16:58] don't get excited [17:17:02] where do you find this stuff [17:17:02] it's pretty confusing [17:17:06] less confusing than vega [17:17:11] oh, Jane told me about it in passing [17:17:20] we were talking about better ways to debug vega [17:18:31] so wait, why can't we use this? [17:20:35] so its not ther competitor - its the same university [17:21:48] and how can i import [17:27:22] milimetric, do you know how to import there? [17:31:05] yurik: in scrum of scrums and other meetings, I'll look in a bit [17:31:25] kafka.server.BrokerTopicMetrics.AllTopicsMessagesInPerSec.FifteenMinuteRate CRITICAL: 7.06787035605e-16 [17:32:16] yurik: weird! they don't have import [17:32:30] milimetric, exactly - they say they do in the readme/notes [17:32:37] mutante: thanks, where is that? (sorry our ops guy is not around) [17:35:50] milimetric: it's on analytics1021 did you meant that? or where i saw the message? that was on icinga web ui [17:36:12] thanks mutante, that's what I meant [17:36:16] hm.... pooop... [17:36:33] jgage: can you help? [17:37:10] milimetric: yw. i just had no idea what it means in practice [17:37:37] mutante: sounds like we're getting less data than we expect maybe [17:38:49] ok, *nod* [17:40:12] mforns: I kee p getting this - ProgrammingError: (_mysql_exceptions.ProgrammingError) (1146, "Table 'wikimetrics_testing.task_error' doesn't exist") [SQL: u'DELETE FROM task_error'] when running tests. Do you have any insight into this error? [17:40:28] madhuvishy, mmmm [17:40:44] madhuvishy, how are your migrations behaving? [17:40:59] madhuvishy, what happens if you execute alembic upgrade head? [17:41:08] milimetric: EL on beta cluster seems to work now (valid and invalid in kafka) [17:41:17] mforns: It complained that there are two heads [17:41:22] hi [17:41:28] hi jgage [17:41:32] mforns: so i ran them one by one [17:41:36] madhuvishy, that is a good candidate [17:41:42] madhuvishy, aha [17:41:59] mforns: do I merge both migrations? [17:42:06] madhuvishy, mmm [17:42:18] madhuvishy, what patch are you trying? [17:42:22] what task? [17:43:33] mforns: I am doing my validate cohort one. I rebased my changes against master - which now has your recent patches [17:43:47] madhuvishy, I see [17:44:09] so before rebasing, you had your validate again migration on top of previous master [17:44:10] mforns: alembic branches now gives me this - [17:44:16] https://www.irccloud.com/pastebin/Y9CQfE3C [17:44:43] madhuvishy: yep, so you have to set 4a2162ae3e84 as the down revision for 35adbe20f3d [17:44:57] mforns: aah [17:45:55] madhuvishy, the normal workflow should be alembic downgrade before rebase, rebase, and then alembic upgrade [17:46:51] madhuvishy, it's confusing.. I've messed up alembic lots of times xD [17:47:15] mforns: ha ha, i foresee messing it up a lot of times [17:48:04] madhuvishy, so now I don't know in which state are your migrations [17:48:16] mutante & milimetrric: ok, analytics1021 problem is fixed. it's a longstanding and mysterious problem we have with this host where its connection between kafka and zookeeper times out and so it stops being a leader for kafka partitions. [17:48:20] gage@analytics1021:~$ kafka topic --describe | grep "Leader: 21" | wc -l [17:48:23] 0 [17:48:25] gage@analytics1021:~$ kafka preferred-replica-election [17:48:28] gage@analytics1021:~$ kafka topic --describe | grep "Leader: 21" | wc -l [17:48:31] 24 [17:48:32] mforns: hmmm let me try setting yours as the downgrade [17:48:39] madhuvishy, and as you rebased, I don't know how you can revert the changes [17:49:03] madhuvishy, what happens if you try alembic downgrade base? [17:49:03] jgage: Weirdddd :) [17:49:05] (no data was lost, we just had a period of 3 active kafka brokers instead of 4) [17:49:17] jgage: thanks!! [17:49:21] :D [17:50:08] the phab task about it: https://phabricator.wikimedia.org/T71667 [17:50:33] https://wikitech.wikimedia.org/wiki/Analytics/Cluster/Kafka/Administration#Replica_Elections [17:50:36] mforns: hmm it seems to be fine - but i think i understand - your migration for the task error table dint show up [17:50:52] i don't mind fixing this when it happens, just sharing the knowledge.. [17:51:00] madhuvishy, aha [17:51:16] mforns: so i have to merge first and then upgrade [17:51:49] madhuvishy, do you need to merge? [17:51:58] madhuvishy, you mean rebase? [17:52:18] mforns: uhhh i dont know. reading alembic docs [17:52:53] madhuvishy, ok [17:53:23] git log [17:53:44] gah i'm typing in the wrong place [17:53:53] madhuvishy, do you want to try solving it in the batcave? [17:53:54] madhuvishy: all i meant is literally go into your revision and type in marcel's revision as your "down" [17:54:09] that's much simpler than doing it "the right way" [17:54:10] :) [17:54:46] milimetric, i will put that ide on wmflabs [17:54:55] jgage: awesome, thanks much [17:54:57] this way ppl can import data [17:55:00] from wiki [17:55:15] yurik: cool, send me the link, I'd like to demo it from there then [17:55:24] I got one night to get good at it :) [17:55:42] milimetric, lets hope we find a way to import :) [17:56:50] milimetric: mforns thanks - think its fixed now [17:57:14] madhuvishy, cool :] [17:58:26] mforns: ok, so the processor seems fine now, and we should pair up on the config [17:58:37] I have a meeting in 30 minutes [17:58:46] milimetric, aha [17:58:50] and I want to eat some lunch, when are you done for the day? [17:59:19] milimetric, today at 22h -> 16h Phili [18:00:06] ok, then I'll ping you as soon as I'm done with my next meeting, which is 15:00 here and 21:00 there [18:00:37] milimetric, mforns : I'll push a review for puppet config tomorrow [18:00:43] for the forwarder [18:00:46] milimetric, ok [18:00:47] joal, ok [18:00:56] Please let me know if you have one already [18:01:20] Done for today ! [18:01:27] Have a good end of day folks [18:02:57] joal|night, good night! [18:17:33] Analytics-Kanban, VisualEditor, Editing Department 2014/15 Q4 blockers: Schema:Edit seems to incorrectly set users as anonymous {lion} - https://phabricator.wikimedia.org/T92596#1282915 (Jdforrester-WMF) Note: Discussed at the weekly Editing triage on 2015-05-13 but not accepted or rejected whilst we f... [19:00:46] mforns: ping :) [19:00:49] to the batcave! [19:02:58] milimetric, going [19:27:56] (PS8) Madhuvishy: WIP: Fix validate again functionality on cohort display page [analytics/wikimetrics] - https://gerrit.wikimedia.org/r/206346 (https://phabricator.wikimedia.org/T78339) [19:27:58] (PS2) Madhuvishy: Fix models to match migrations [analytics/wikimetrics] - https://gerrit.wikimedia.org/r/206343 (owner: Milimetric) [19:34:01] mforns/milimetric: what's the staging server called? [19:34:14] for wikimetrics [19:34:23] wikimetrics-staging1.eqiad.wmflabs [19:34:35] cool thanks [19:59:30] mforns: you'll love what it was doing [19:59:38] milimetric, ? [19:59:42] instead of pasting, it was trying to paste 98779 times [19:59:48] because it was interpreting what I pasted as commands [19:59:49] milimetric, xDDD [20:00:26] i opened the recovered file and it was full of 98779 copies of the bug number, so funny :) [20:00:27] milimetric, again: vi is for crazy people :] [20:00:34] xD [20:04:50] :) I rarely screw up anymore, but when I do, I do it big [20:05:11] bye everyone! see you tomorrow [20:05:13] it's mostly because of the inconsistencies, the "gui" version and the "cli" version never agree on settings [20:05:15] nite! [20:05:30] milimetric, aha [20:05:54] g'night [21:08:16] milimetric, didn't have enough time to set up the lily. If you can, could you install it somewhere? possibly on a separate instance? [21:08:32] i'm off to bed, dead tired, and will need to make a presentation tomorrow morning [21:09:24] I'll give it a shot, yurik, good night [21:09:48] milimetric, thx! make sure you add me as an admin there :) [21:09:58] i mean - install it in a group that has both of us as admins [21:10:13] milimetric: tested my change on staging - it seems to work fine! [21:10:45] milimetric: although my css change is not showing up. [21:12:04] madhuvishy: that's great [21:12:12] CSS might be a caching issue - did you clear? [21:12:20] hm, no, but we have cache busting [21:12:35] milimetric: its not even CSS - its html [21:13:38] madhuvishy: i see the validate again button, what html change? [21:15:01] milimetric: hmmm i spaced the buttons. i think the code is not pulled right. i dont see it on the staging server code. checking what happened [21:15:33] yeah, if it's html you don't even have to restart anything [21:15:58] how'd you grab the code? The easiest way is sudo-ing the command that gerrit lets you copy/paste for checkout [21:16:14] and use the anonymous http version [21:18:20] milimetric: aah. that's what i was missing. keep getting permission denied [21:26:58] milimetric: hmmm i think it's fine now [21:27:20] great work. I have to look at it in a bit or tomorrow because I have a meeting coming up [21:27:35] but if you want to look at the puppet stuff I did in a half hour or so, I can do that [21:27:52] milimetric: yeah no problem [21:27:57] sure we can do that [22:42:18] (PS3) Madhuvishy: Add generic oozie component for emailing data [analytics/refinery] - https://gerrit.wikimedia.org/r/210632 [22:46:53] madhuvishy: ok, free [22:48:07] milimetric: cool. let me just get some water. then batcave? [22:49:46] Analytics-Tech-community-metrics, ECT-May-2015: Provide list of open Gerrit changesets with most activity which aren't -1/-2'ed - https://phabricator.wikimedia.org/T94036#1283775 (Dicortazar) Hi again, I already fixed the query to get the results. This is the top 10 (well, with the last version I've loc... [22:50:03] to the batcave! [23:19:34] Analytics-Kanban, VisualEditor, Editing Department 2014/15 Q4 blockers: Schema:Edit seems to incorrectly set users as anonymous {lion} - https://phabricator.wikimedia.org/T92596#1283855 (Halfak) Open>Resolved a:Halfak Upon re-review with DAndreescu, it looks like this field is set appropriatel... [23:42:13] Analytics-Kanban, VisualEditor, Editing Department 2014/15 Q4 blockers: Schema:Edit seems to incorrectly set users as anonymous {lion} - https://phabricator.wikimedia.org/T92596#1283891 (ggellerman) Thanks, @Halfak!