[12:06:07] (PS1) QChris: Add toolbelt with Info tool [analytics/kraken] - https://gerrit.wikimedia.org/r/115888 [12:06:09] (PS1) QChris: Add Store to toolbelt [analytics/kraken] - https://gerrit.wikimedia.org/r/115889 [12:41:01] (PS2) QChris: Add Store tool to toolbelt [analytics/kraken] - https://gerrit.wikimedia.org/r/115889 [12:41:03] (PS1) QChris: Add Dump tool to toolbelt [analytics/kraken] - https://gerrit.wikimedia.org/r/115892 [13:47:45] qchris_away: hi [13:48:01] qchris_away: in order to have tests for these queries.. we'd need a hive in local mode on jenkins [13:48:25] hashar_: hey Antoine, do you think we could have a hive just for local mode in jenkins ? [13:48:46] average: what do you mean? [13:49:02] hashar_: let me explain. we have these queries, that are supposed to run on Hive [13:49:08] ahhh [13:49:12] hashar_: but jenkins shouldn't mess with production Hive [13:49:15] dont you mock them up in your tests ? [13:49:18] hashar_: kindof yes [13:49:19] yes [13:49:24] no no [13:49:25] wait [13:49:30] end of story :-] [13:49:41] so, the "mocking" in this case could be done with a local Hive [13:49:46] so for unit testing, you should probably not need any Hive [13:49:59] I could generate like fake data and throw it in a local Hive, and run the queries there, on jenkins, with fake data [13:50:23] well, I kinda need it.. but the good news is it will only need to run in local mode [13:51:06] we could probably get hive installed on labs somehow [13:51:10] hashar: in short. I just need a hive and a hadoop installed or placed on jenkins. http://garage-coding.com/releases/udfs-v1/README.txt (see the "Hive in Local Mode" section) [13:51:14] did that for elastic search browser tests [13:51:19] hashar: but labs is not jenkins [13:51:33] jenkins now has slaves running on labs instances :] [13:52:04] so we can get hive/hadoop installed there probably [13:52:15] but if you install it, it'll be overkill [13:52:18] let me explain why [13:52:41] if you install it on the labs nodes, you'll get some processes hanging around and they won't be used. Hive in local mode doesn't need Hadoop running per-se.. [13:52:47] it only needs the hadoop binaries [13:52:53] it will use the regular filesystem [13:52:56] no hdfs [13:54:27] ottomata: did you manage to upgrade to 0.8.3 yesterday? [13:56:45] hashar: ok, I guess it's way to complicated, nevermind [13:56:51] s/to/too/ [13:59:34] average: I have no idea what hdfs/hive/hadoop are :-] [13:59:46] average: so if you need one of those installed, we can surely get ti done [14:00:00] average: or have your tests run on a dedicated Jenkins slave that you would maintain. Parsoid does that [14:01:44] Ok, here's a more simplified version: I only need 2 zip files unzipped and sitting at a certain location where the jenkins job has read/write rights. [14:01:51] is that possible ? [14:19:41] Snaps: hiayyy, I upgraded it on cp3019 only [14:34:55] average: I am working on getting it done without changing the Jenkins hosts. Just "mvn test" and you're good [14:35:01] No hadoop installed. [14:35:04] No hive installed. [14:35:15] wow [14:35:23] that sounds amazing qchris [14:35:24] Oh ... average is not around ... [14:35:33] it's ok, i'm listening [14:35:34] milimetric: It's not yet fully done :-) [14:35:36] you're not talking to yourself [14:35:39] Haha. [14:35:40] Yes. [14:35:43] i didn't say *is* amazing :) [14:36:15] ;-) [15:07:07] (PS16) Stefan.petrea: [WIP] kraken-hive UDFs [analytics/kraken] - https://gerrit.wikimedia.org/r/96738 (owner: QChris) [15:09:30] hashar: how can I tell zuul not to run stuff for a specific patchset ? [15:09:51] hashar: to not run tests for a specific patchset. because I pushed partial work, and there was no point to run the tests on it [15:17:48] average: ignore the test results until the patch is completed ? :D [15:59:34] (CR) Ottomata: "Stefan, I just started reviewing the first few files and noticed that some of Christian's comments were not addressed in the most recent p" [analytics/kraken] - https://gerrit.wikimedia.org/r/96738 (owner: QChris) [18:37:10] (PS14) Nuria: [WIP] Changes tu support wikimetrics in vagrant. [analytics/wikimetrics] - https://gerrit.wikimedia.org/r/109676 [19:27:41] (CR) Milimetric: [WIP] Changes tu support wikimetrics in vagrant. (3 comments) [analytics/wikimetrics] - https://gerrit.wikimedia.org/r/109676 (owner: Nuria) [21:04:53] (PS15) Nuria: [WIP] Changes tu support wikimetrics in vagrant. [analytics/wikimetrics] - https://gerrit.wikimedia.org/r/109676 [21:05:06] sweet patchy goodness :) [21:05:55] sorry forgot to add 1 file.....arghhh [21:06:21] (PS16) Nuria: [WIP] Changes tu support wikimetrics in vagrant. [analytics/wikimetrics] - https://gerrit.wikimedia.org/r/109676 [21:06:47] ok... now for real [21:24:52] hrm. do we have the C API for MaxMind, or just the db itself? [21:25:45] (PS17) Milimetric: [WIP] Changes tu support wikimetrics in vagrant. [analytics/wikimetrics] - https://gerrit.wikimedia.org/r/109676 (owner: Nuria) [21:27:30] I think we have the C API [21:27:32] no? [21:27:35] hm not sure [21:28:46] ottomata, context; if we want to be doing geolookup for non-requestlog data (say, the recentchanges feeds) on an ad-hoc basis, poking /usr/bin/geoiplookup or whatever via the terminal is sloooow. [21:29:00] well, unless you want to write exclusively in bash [21:29:04] so I'm exploring ways to speed it up. [21:32:39] libgeoip1 is installed on stat1002 [21:32:46] Version: 1.4.8+dfsg-2 [21:33:35] [21:33:36] /usr/lib/libGeoIP.so.1 [21:33:58] aha; cool! [21:33:58] thanks :)