[09:59:01] hey hashar [09:59:24] i am setting up jenkins for kraken but i am missing a couple of dependencies [09:59:26] see https://integration.wikimedia.org/ci/view/Analytics/job/Kraken/3/ [10:00:02] the libgeoip databases need to be installed on gallium for example [10:00:27] the libdclasslib should be installed as well (i belief that one is in apt.wikimedia.org) [10:01:38] drdee: hey [10:01:47] hey [10:02:00] libgeoip-dev is installed :) [10:02:25] that is done by having puppet include geoip [10:03:01] I can't find any libdclasslib package in apt :( [10:03:08] typo ? [10:03:08] :)D [10:03:41] about lbgeo-ip: it also needs all the geoip databases in /usr/share/geoip [10:03:54] libdclasslib: i am searching in RT [10:03:54] 1 sec [10:04:13] ls: cannot access /usr/share/geoip: No such file or directory [10:04:13] :D [10:04:20] maybe it can not be installed on that box [10:06:07] http://garage-coding.com/releases/libdclass-dev/ it hasn't been put in apt.wikimedia.org yet [10:06:09] but it should [10:06:24] that's the private site of stefan, our external contractor [10:06:33] can you put it in apt.wikimedia.org ? [10:06:38] nop [10:06:42] who can? [10:06:45] is that a [10:06:47] sorry [10:06:56] what is that lib ? is that maintained by some upstream or just stefan ? [10:07:05] it's maintained by us [10:07:15] is that in Gerrit ? :-] [10:07:29] iirc the ops process is to get a repo under operations/debs/ [10:07:41] any changes get reviewed by ops / merged [10:07:47] and then they build the package / upload it on apt [10:07:48] right [10:08:02] k [10:08:03] ottomatta can probably guide you / get it done. [10:08:20] so are there any geoip databases installed on gallium? [10:08:22] I myself don't have write access on ops/* repos nor I got access on apt.wm.o [10:08:28] k [10:08:36] if not, can you install those db's on gallium? [10:08:58] the GeoIP are under /usr/share/GeoIP/ [10:09:12] which db's are? [10:09:32] GeoIPCity.dat GeoIP.dat GeoIPRegion.dat GeoIPv6.dat [10:09:36] awesome [10:09:43] 2e326739466805abc3b0409545b21482 GeoIPCity.dat [10:09:44] 6f7aa9d67f496c6a74c7e8f9bb71a98f GeoIP.dat [10:09:45] 688d8536bcac041ae601f0d65fe48545 GeoIPRegion.dat [10:09:45] c7367aac1a04192842de200f5b41ab76 GeoIPv6.dat [10:09:49] I guess the ones from production :-] [10:10:23] finally, jenkins can't find http://repo.maven.apache.org/maven2/storm/storm/0.8.2/storm-0.8.2.pom [10:10:49] that is a 404 :-] [10:12:58] brb nature's call + cigarettes [10:13:30] http://clojars.org/repo is the right maven repo [10:22:12] back [10:33:55] gotta run but http://clojars.org/repo is the repo that contains the storm dependencies [10:34:01] can you add that to jenkins? [10:34:22] drdee: I have no idea how that is to be done [10:34:27] aight [10:34:28] I guess it should be defined in the maven file [10:34:30] sorry :( [10:34:33] i will look into it [10:41:04] I am out for lunch [12:29:29] I would be very interested to see the typical cpu usage on a typical frontend machine under load. How much cpu is varnish consum, using how many cores, etc.. [12:29:48] s,consum,consuming, [13:26:04] moorning [13:54:57] hey sorry i'm late - IRC is fubar after 13.04 upgrade [14:57:39] heya, drdee, you there? [15:15:14] yes [15:15:23] ottomata ^^ [15:16:41] i'm getting an index out of bounds exception for the pageview udf [15:16:56] Caught error from UDF: org.wikimedia.analytics.kraken.pig.PageViewEvalFunc, Out of bounds access [-1] [15:17:06] can you paste the stack trace ? [15:17:07] Caused by: java.lang.ArrayIndexOutOfBoundsException: -1 [15:17:07] at org.wikimedia.analytics.kraken.pageview.ProjectInfo.(ProjectInfo.java:344) [15:17:10] http://localhost:19888/jobhistory/logs/analytics1019:8041/container_1369058864829_0256_01_000004/attempt_1369058864829_0256_m_000002_0/otto [15:17:18] ok, will look at it [15:17:26] hey milimetric! [15:17:37] howdy [15:24:03] oh guys, i moved this weeks scrum related meetings one hour later [15:24:23] oh, just noticed that, hm ok [15:27:35] so you can have lunch :) [15:27:43] (and I can have dinner) [15:28:19] What is for dinner? [15:28:27] steak [15:28:31] nice [15:28:41] every day steak [15:28:56] are you able to join the UMAPI meeting with Frank? [15:29:24] yes [15:35:45] ok, drdee, i *think* the new zero country stuff is working, but I can't tell without the pageview thing fixed, and i'm not really sure what to do from here [15:39:22] i will fix that [15:39:30] talking to adrian now [16:06:26] milimetric; any news on #88 [16:16:24] ottomata: will push fix soon, trivial bug [16:21:09] david looked at #88 over the last couple of days [16:21:13] drdee^ [16:21:32] he said that he thought my iterator should work, and can't see the problem [16:21:38] but that he's looking into it more [16:21:42] k [16:22:11] I'm currently trying to figure out how the mobile platform job is set up so i can update it - tomasz said it's ok to get all the results in one file [16:27:57] cool [16:32:39] hey ottomata: i heard you and erosen were looking into hue logins not working [16:33:22] oh yeahhhhhh not sure what is up with that…mine is working [16:49:04] so what username do you use to login to hue ottomata? [16:50:13] otto [16:50:15] my shell username [16:51:29] hm... but... what password then? same as labs? [16:52:56] when i use milimetric with my labs password, it says invalid username or password [16:53:25] i just thought this was fixed when i got renamed [17:03:46] same as labs [17:05:48] ok, that's not working [17:06:03] neither is any of the million combinations i tried [17:06:27] so how do i figure out where in hadoop the oozie coordinator and properties are for the mobile platform job? [17:06:36] we really need to fix dan's access to hue [17:06:59] he is wasting way too much time on these trivial things [17:09:55] ok, i'm about to commit a monitoring thingee [17:09:57] will see what I can do [17:09:58] amen^2 [17:10:05] thx ottomata [17:10:27] btw ottomata, please have a more in-depth look into apache bigtop; it has a lot of puppet and debian stuff for hadoop and hadoop ecosystem [17:10:58] iiinteretsing [17:11:03] diner time, scrum in 50 minutes! [17:34:44] woohooo [17:34:44] https://icinga.wikimedia.org/icinga/ [17:34:54] ok, looking at hue now... [17:36:35] milimetric: try logging in [17:36:45] k [17:37:27] now it's just sitting there spinning [17:37:37] am i supposed to have another ktunnel besides 8888 up? [17:38:13] hmmmm, no [17:38:21] ok hang on [17:56:08] arrgh [17:56:19] my cat just demolished my keyboard [17:59:54] haha [18:00:02] (milimetric, i thikn i got it, few more mintes...) [18:02:20] scrum [18:02:26] ottomata: ^^ [18:38:53] oh erosen, i've been thinking, and i'm now leaning towards leaving amsterdam on sunday eve [18:39:01] k [18:39:04] not quiiite sure yet, but i need to by train ticket soon [18:39:27] well, I haven't booked a place yet, so it should be fine either way [18:39:33] just let me know when you decide [18:39:42] k [18:39:43] mmk [18:40:11] milimetric: http://gp.wmflabs.org/graphs/grants_count_by_global_south [18:51:50] hah, welp yay the new kafka alert is working, i just got it! something is busted again…but not in the same way [18:52:19] can i also get those alerts? [18:54:41] milimetric: try logging into hue now [18:54:51] k, doing [18:54:59] you should get them drdee [18:55:03] its to the analytics contactgroup [18:55:08] k [18:55:11] yay [18:55:23] erosen: you too [18:55:25] try logging into hue [18:55:35] k [18:57:22] yay [18:57:30] ottomata: worked [18:57:38] coooool [18:57:43] thanks to mark! [18:58:16] heh, that jar - hell thing is pretty bad [18:58:30] i just realized that david changed the mobile-platform job to use some other jars [18:58:35] i had compiled special ones for it [18:59:55] i just copied the new jars to an10 [19:00:09] but milimetric don't want to mess up your work either [19:00:13] do we need to coordinate? [19:00:24] maybe [19:00:28] ok, so did you build getting latest? [19:00:58] so, oozie.libpath in my job is set at ${nameNode}/libs/kraken-0.0.2 [19:01:37] drdee: umapi meeting? [19:01:40] yup [19:01:43] milimetric: ^^ [19:01:56] yep, ocming [19:20:02] ok, kafka looking better, hue working, job ready but waiting on .jar decision [19:20:11] i'm going to run home and pack a bit before the meeting at in 40 [19:20:14] back on in a bit [20:04:40] i'm in hangout [20:04:46] drdee? [20:04:49] ottomata: wrapping up umapi meeting [20:04:54] coming [20:05:05] k [20:13:07] on my way [20:21:22] erosen: joining us for grooming? [21:37:18] hm, ottomata, that doesn't seem to work, the PIG storage overwrites every time [21:37:21] looking it up [21:40:49] i'm trying to instead of storing into $output, storing into $output/official and $output/unofficial [21:41:57] hmm, yeah [21:42:01] hmm [21:42:11] i guess that's ok [21:42:15] hmm [21:43:10] ok, cool, so that works [21:43:25] now will the job that concats and sorts these go recurse into the directories? [21:44:15] hm, not sure [21:44:16] i think so [21:44:30] ok, we'll chance it :) [21:44:35] yeah i htink so [21:49:35] ok ottomata, all set - hdfs_sync.sh is run [21:49:53] milimetric: do you need my jars? [21:50:02] i thought those were deployed [21:50:22] they were copied to an10 [21:51:11] copy them to /libs ? [21:52:27] to /libs/kraken right? [21:52:27] if so, then we're all set [21:52:27] now i'm looking at the coordinator in oozie [21:52:27] and should i hit the kill button? then i'm assuming it'll let me rerun [21:52:28] or should I just let you do that ottomata? [21:52:40] so wait the jars are on an10 in some temporary folder drdee? [21:52:49] or in /libs/kraken? [21:52:50] in my home folder [21:52:53] on an10 [21:52:54] oh! [21:53:11] i think both andrew and i thought you had deployed them officially [21:53:17] so when were you going to do that? [21:53:24] or is it not ready? [21:53:28] it's ready [21:53:32] just was waiting for the go [21:53:38] do it now [21:53:38] brb, shower [21:53:38] ? [21:54:11] but i am confused, where to put them? [21:54:17] /libs ? [21:54:22] yes [21:54:27] /libs/kraken-0.0.2 [21:54:33] /libs/kraken? [21:54:49] oh ok, one sec [21:55:59] this is what's confusing me too drdee: [21:56:01] krakenLibPath hdfs://analytics1010.eqiad.wmnet:8020/libs/kraken [21:56:12] so that job is set up to look in /libs/kraken [21:56:18] in hfds [21:56:27] but all the jars are in /libs [21:56:52] /libs/kraken/ contains two folders pig & oozie [21:56:54] that's all [21:57:18] and pig contains all the pig scripts [21:57:47] maybe we need //libs/kraken/jars/ [21:58:04] and move all the jars over there [21:58:15] yeah, no idea where these are supposed to be to be read properly [21:58:31] i see the oozie lib-path has /libs/kraken-0.0.2 [21:58:42] aararghhh [21:59:05] and it looks like david copied a bunch of stuff into there [21:59:10] that might be the place to put them... [21:59:34] yeah, i think that's right [21:59:39] a few other jobs reference that location [21:59:59] ottomata can tell you if that's where he's expecting them as far as he knows (when he's out of the shower) [22:00:20] we need to clean this up [22:00:24] i've gotta run but i'll be checking email if there's anything urgent [22:00:26] totally agree [22:01:06] for now, i'd just deploy them to /libs/kraken-0.0.2 and maybe move the current /libs/kraken-0.0.2 to a backup folder [22:01:48] once that's done, if you don't mind telling Andrew to re-run the mobile_platform job that'd be cool [22:01:58] ok [22:04:06] done [22:04:13] new jars are in /libs/kraken-0.0.2/ [22:04:27] ohhh didn't' backup [22:04:44] sorry [22:04:46] :( [22:15:30] hey guys [22:15:33] sorry, how goes it? [22:18:10] drdee, milimetric^ [22:19:14] ottomata: can't help much but you might want to look over the log [22:19:30] milimetric: wanted to ask you to rerun his job [22:21:39] ottomata: did you end up re-running the country counts job? [22:21:43] after refactoring [22:22:01] no, was waiting for the jar fix that drdee just uploaded [22:22:06] gotcha [22:22:09] cool [22:22:25] did they sign off? [22:22:29] its pretty late for drdee i guess [22:22:34] yeah, not sur [22:22:40] i think dan might be back [22:22:45] but was gonna be out for a while