[11:06:08] <ottomata>	 qchris, i'm moving to a cafe
[11:06:15] <ottomata>	 but check out what nuria and I have been working on!
[11:06:20] <ottomata>	 http://etherpad.wikimedia.org/p/analytics-naming
[11:06:28] * qchris  checks it out

[11:12:54] <qchris>	 ottomata: Does that mean we again have oozie files and sources for jars in the same repository (i.e.: furnace)?
[11:13:50] <qchris>	 (or should line 80 by furnace-deploy/config?)
[11:16:14] <ottomata>	 furnace/config is a different repository
[11:16:35] <qchris>	 Ah. ok.
[11:16:57] <ottomata>	 i'm not 100% for that, but nuria convinced
[11:17:06] <ottomata>	 me
[11:17:08] <ottomata>	 and you too!
[11:17:15] <ottomata>	 ok, heading to cafe, back in 20ish prob
[11:17:22] <qchris>	 Ok.
[11:20:31] <nuria>	 boy , we spent some time eh?
[11:20:39] <nuria>	 it is already 1pm?
[11:33:21] <grrrit-wm>	 (CR) Nuria: "Thanks much for doing changes! Could we rephrase the asserts a bit?" (2 comments) [analytics/wikimetrics] - https://gerrit.wikimedia.org/r/131232 (https://bugzilla.wikimedia.org/64724) (owner: Terrrydactyl)
[11:56:57] <ottomata>	 back!
[11:58:56] <grrrit-wm>	 (PS1) Gilles: Generate SQL to a separate folder [analytics/multimedia] - https://gerrit.wikimedia.org/r/133223
[13:11:16] <grrrit-wm>	 (PS1) Ottomata: Now importing all webrequest data into a webrequest/ subdirectory [analytics/kraken] - https://gerrit.wikimedia.org/r/133229
[13:11:30] <grrrit-wm>	 (PS2) Ottomata: Now importing all webrequest data into a webrequest/ subdirectory [analytics/kraken] - https://gerrit.wikimedia.org/r/133229
[13:11:35] <grrrit-wm>	 (CR) Ottomata: [C: 2 V: 2] Now importing all webrequest data into a webrequest/ subdirectory [analytics/kraken] - https://gerrit.wikimedia.org/r/133229 (owner: Ottomata)
[13:12:01] <grrrit-wm>	 (PS1) Ottomata: Updating kraken [analytics/kraken/deploy] - https://gerrit.wikimedia.org/r/133230
[13:12:15] <grrrit-wm>	 (PS5) Ottomata: Adding oozie files to control addition and deletion of webrequest table partitions [analytics/kraken] - https://gerrit.wikimedia.org/r/131208
[13:12:20] <grrrit-wm>	 (CR) Ottomata: [C: 2 V: 2] Adding oozie files to control addition and deletion of webrequest table partitions [analytics/kraken] - https://gerrit.wikimedia.org/r/131208 (owner: Ottomata)
[13:12:58] <grrrit-wm>	 (PS2) Ottomata: Updating kraken with new oozie files and camus webrequest config for single table [analytics/kraken/deploy] - https://gerrit.wikimedia.org/r/133230
[13:13:11] <grrrit-wm>	 (CR) Ottomata: [C: 2 V: 2] Updating kraken with new oozie files and camus webrequest config for single table [analytics/kraken/deploy] - https://gerrit.wikimedia.org/r/133230 (owner: Ottomata)
[13:19:34] <halfak>	 ottomata, I just wanted to let you know that I'm stoked to follow your emails and progress WRT hadoop.
[13:20:00] <halfak>	 I'm jealous though that we're getting a single table solution for request before we got one for revisions though. :P
[13:20:21] <ottomata>	 :)  help toby/kevin prioritize that stuff
[13:20:35] <ottomata>	 i'm hardly thinking about anything other than webrequest
[13:20:42] <ottomata>	 because it is the big big data :)
[13:20:47] <ottomata>	 that there is no other way to process
[13:20:57] <ottomata>	 but sqooping up stuff from mysql into hive *should* be relatively easy
[13:21:01] <halfak>	 No worries.  Webrequest is high priority now because I can work within the constraints of split language DBs.
[13:38:57] <ottomata>	 woot, partitions are rolling in for new webrequest table! (not via oozie)
[13:39:00] <ottomata>	 oozie soon!
[13:39:21] <ottomata>	 select count(*) from webrequest where webrequest_source='bits' and year=2014 and month=05 and day=01 and hour=01;
[13:39:24] <ottomata>	 85384284
[13:39:34] <halfak>	 ^ +1
[13:51:03] <Krinkle>	 Is something up with analytics / event logging?
[13:51:09] <Krinkle>	 Looks like graphite data is broken as of 24 hours
[13:51:22] <Krinkle>	 https://graphite.wikimedia.org/render/?target=mw.js.deprecate.addOnloadHook.rate&from=-12hours
[13:51:32] <Krinkle>	 random spots come in here and there, but the overall line gone
[13:51:53] <Krinkle>	 https://graphite.wikimedia.org/render/?target=mw.js.deprecate.addOnloadHook.rate&from=-3days
[13:52:17] <halfak>	 Doesn't look like eventlogging is generally down.
[13:52:29] <halfak>	 I just checked on db1047 and I'm getting recent timestamps.
[13:52:36] <Krinkle>	 Looked at various other properties, same thing
[13:53:03] <halfak>	 Could be some other part of the pipeline.
[13:53:15] <Krinkle>	 Maybe the statsd deamon or the graphite receiver is overloaded? It's like it's only recording spots at random intervals with large gaps
[13:53:41] <halfak>	 Oh gaps.  Let me look for that.
[13:54:45] <halfak>	 No gaps in ServerSideAccountCreation
[13:54:50] <halfak>	 (looking hourly)
[13:54:59] <halfak>	 Is there a specific schema that you are looking at Krinkle?
[13:55:23] <Krinkle>	 mw.js.deprecate ; webperf/deprecate.py / Schema:DeprecatedUsage
[13:56:41] <halfak>	 Yeah... no meaningful dips there either.  I see a seasonal trend where count goes down to ~1200 in the wee hours and back up to ~3000 around 8AM UTC.
[13:56:57] <Krinkle>	 How are you seeing that?
[13:57:09] <halfak>	 Querying the EventLogging tables in the DB directly.
[13:57:19] <halfak>	 I'll pastebin in a sec.
[13:57:59] <halfak>	 http://pastebin.com/4Uixp3sy
[13:58:48] <Krinkle>	 halfak: http://codepen.io/Krinkle/full/KjhIf/
[13:59:28] <halfak>	 Yup.  Definitely looks like a bunch of graphs with recent data strangeness.
[13:59:34] <Krinkle>	 a lot of them have rather strange anomalies. Nothing like anything in the past 3 weeks.
[14:00:12] <halfak>	 Agreed.  I think it is somewhere after EventLogging in the pipeline leading to those graphs.
[14:00:30] <Krinkle>	 Can you do that query again for an individual property?
[14:00:30] <halfak>	 Could it be that the cron script generating counts was having issues?
[14:00:39] <halfak>	 Sure!
[14:00:48] <Krinkle>	 e.g.this one http://graphite.wikimedia.org/render/?title=appendCSS&width=300&height=200&target=alias(mw.js.deprecate.appendCSS.rate%2C%22rate%22)&from=-3days
[14:00:52] <Krinkle>	 method=appendCSS
[14:01:58] <halfak>	 Looks good http://pastebin.com/21j521Jq
[14:05:40] <Krinkle>	 are you familiar with the infrastructure between eventlogging database and graphite?
[14:05:51] <Krinkle>	 (I'm not, other than knowing what the different components are called)
[14:06:06] <halfak>	 Sadly no.  I'm not even sure who would be.
[14:06:23] <halfak>	 milimetric might know (ping!)
[14:07:01] <milimetric>	 halfak / Krinkle I'm in daily standup now
[14:07:11] <milimetric>	 but we talked about this and we'll look at it soon
[14:07:22] <halfak>	 <3
[14:07:23] <halfak>	 thanks
[15:10:22] <qchris>	 Krinkle, halfak: I just checked gerrit metrics in graphite,
[15:10:33] <qchris>	 and they also break around the same time and same date.
[15:10:51] <qchris>	 So maybe it's some problem in the parts that feed graphite?
[15:10:58] <qchris>	 (Thinking statsd and such)
[15:11:36] <milimetric>	 I second qchris's opinion ^
[15:11:50] <qchris>	 Is this a bugzilla thing, or an RT thing?
[15:13:37] <Krinkle>	 ?
[15:13:49] <Krinkle>	 Oh, I see what you mean. Where to file the bug?
[15:13:58] <qchris>	 Yup.
[15:14:02] <Krinkle>	 I don't know. Does any of you have access to fix it?
[15:14:07] <Krinkle>	 Do
[15:14:09] <qchris>	 mili metric is gonna ask in ops.
[15:14:12] <Krinkle>	 k :)
[15:14:28] <qchris>	 no access to fix that for qchris :-)
[15:16:04] <milimetric>	 yeah, we have no access to fix this stuff but we suspect ori and chase were looking at it yesterday from some IRC logs we saw
[15:22:07] <milimetric>	 Krinkle: see in operations channel, chase is looking at it
[15:43:12] <grrrit-wm>	 (PS1) Ottomata: Initial repository layout [analytics/refinery/source] - https://gerrit.wikimedia.org/r/133257
[15:53:56] <ottomata>	 grr, i can't get to gerrit to push the analytics/refinery layout commit
[15:53:57] <ottomata>	 and I gotta run
[15:54:04] <ottomata>	 i guess no hurry, will push tomorrow
[15:54:05] <ottomata>	 byebyeyyy
[17:00:48] <grrrit-wm>	 (PS1) QChris: Switch back to normal slave for s1 again [analytics/geowiki] - https://gerrit.wikimedia.org/r/133273
[17:02:20] <grrrit-wm>	 (PS5) Terrrydactyl: Fixes problem with deleting empty cohorts. [analytics/wikimetrics] - https://gerrit.wikimedia.org/r/131232 (https://bugzilla.wikimedia.org/64724)
[17:43:02] <grrrit-wm>	 (PS2) QChris: Switch back to normal slave for s1 again [analytics/geowiki] - https://gerrit.wikimedia.org/r/133273
[18:53:06] <grrrit-wm>	 (PS2) Terrrydactyl: [WIP] Add ability to tag a cohort [analytics/wikimetrics] - https://gerrit.wikimedia.org/r/133091
[18:55:32] <Ironholds>	 okay; milimetric, you look like you might be awake.
[18:55:35] <Ironholds>	 Can I ask for a favour?
[19:02:16] <milimetric>	 hi Ironholds
[19:02:20] <grrrit-wm>	 (CR) QChris: [C: -1] "I thought the layout within the source directory from" [analytics/refinery/source] - https://gerrit.wikimedia.org/r/133257 (owner: Ottomata)
[19:02:30] <milimetric>	 I'm sadly late for a meeting
[19:02:36] <milimetric>	 but yes - awake and how can I help?
[19:02:57] <Ironholds>	 I may be able to get the dependencies for RJDBC - the hive-to-R package - set up on my own
[19:03:27] <Ironholds>	 with one exception which is running R CMD javareconf on stat2 as root to ensure R is properly registering where the JDK lives.
[19:03:35] <Ironholds>	 any chance you could run that?
[19:03:47] <milimetric>	 oh no, sadly, I don't have root
[19:04:46] <milimetric>	 but send ottomata an email so he can see it tomorrow morning
[19:04:51] <milimetric>	 Ironholds: ^
[19:44:16] <grrrit-wm>	 (CR) MarkTraceur: [C: 2 V: 2] "Let's see how this goes!" [analytics/multimedia] - https://gerrit.wikimedia.org/r/133223 (owner: Gilles)
[20:00:35] <Ironholds>	 milimetric, cool! Thanks :)
[20:18:39] <grrrit-wm>	 (PS1) MarkTraceur: Fix deploy and generation scripts to be nicer [analytics/multimedia] - https://gerrit.wikimedia.org/r/133353
[20:18:49] <grrrit-wm>	 (CR) MarkTraceur: [C: 2 V: 2] Fix deploy and generation scripts to be nicer [analytics/multimedia] - https://gerrit.wikimedia.org/r/133353 (owner: MarkTraceur)
[20:19:04] <YuviPanda>	 marktraceur: so much V+2
[20:20:07] <marktraceur>	 YuviPanda: No jenkins in that repo yet
[21:04:43] <hashar>	 marktraceur: https://www.mediawiki.org/wiki/Continuous_integration/Tutorials/Test_your_python :-D
[21:04:58] <hashar>	 gotta copy paste some content to a tox.ini file
[21:05:02] <marktraceur>	 Cool!
[21:05:06] <hashar>	 then define a bunch of jobs using a very basic template
[21:05:14] <hashar>	 (copy paste from the page as well)
[21:05:17] <marktraceur>	 *nod*
[21:05:18] <hashar>	 then Zuul triggers, done!
[21:05:52] <hashar>	 I am too lazy for it, but feel free to ask folks to use that tutorial to add tests for their python based repos
[21:13:30] <marktraceur>	 We actually just added a python script to analytics/multimedia
[21:13:31] <marktraceur>	 So
[21:42:30] <grrrit-wm>	 (PS1) Milimetric: Update with wikidata data [analytics/reportcard/data] - https://gerrit.wikimedia.org/r/133367
[21:42:42] <grrrit-wm>	 (CR) Milimetric: [C: 2 V: 2] Update with wikidata data [analytics/reportcard/data] - https://gerrit.wikimedia.org/r/133367 (owner: Milimetric)
[21:54:50] <grrrit-wm>	 (PS2) Milimetric: Use MariaDB 10 sequence table instead of information_schema.columns. Touching i_s.columns means the table is regenerated every time which hammers table cache and file handles. All-shards slaves really suffer. [analytics/limn-mobile-data] - https://gerrit.wikimedia.org/r/131929 (owner: Springle)
[21:57:36] <grrrit-wm>	 (CR) Milimetric: [C: 2 V: 2] "I checked that this approach would work, but since I can't run the job on stat1003, I'll keep an eye out so it doesn't blow up the dashboa" [analytics/limn-mobile-data] - https://gerrit.wikimedia.org/r/131929 (owner: Springle)