[05:27:58] New patchset: Rfaulk; "add. method parse_request to parse user metrics request components." [analytics/user-metrics] (repair_runtime) - https://gerrit.wikimedia.org/r/74111 [05:27:58] New patchset: Rfaulk; "fix. typo / pep8." [analytics/user-metrics] (repair_runtime) - https://gerrit.wikimedia.org/r/74112 [05:28:18] Change merged: Rfaulk; [analytics/user-metrics] (repair_runtime) - https://gerrit.wikimedia.org/r/74111 [05:28:31] Change merged: Rfaulk; [analytics/user-metrics] (repair_runtime) - https://gerrit.wikimedia.org/r/74112 [13:42:42] ottomata: hi [13:42:58] ottomata: would you like to review some packages please ? [13:43:59] hiyaa [13:44:03] yesssuh i would love to! [13:44:15] ottomata: https://gerrit.wikimedia.org/r/#/c/73860/ [13:44:19] drdee: hi [13:45:00] hey average [13:45:03] ok great, i will build it and give comments [13:45:20] ottomata: well, this is just for our upstream [13:45:20] i'm not going to be able to comment much on the build system stuff that faidon did/wants you to do, but I'll see what I find about the debian/ side of things [13:45:23] k [13:45:27] oh [13:45:27] oh [13:45:28] i see [13:45:28] oh [13:45:31] hm [13:45:32] j [13:45:32] k [13:45:33] ottomata: there will be a new patchset for the debian/* [14:07:06] ottomata: oracle jdk 6 necessary [14:07:11] ottomata: it will provide the jni.h [14:07:13] ottomata: https://launchpad.net/~webupd8team/+archive/java [14:08:54] he qchris [14:09:02] Hi drdee :-) [14:09:09] oh i am in openjdk, ok will switch [14:10:58] ottomata, can you give qchris access to stat1002? your ticket was opened on monday, today it's wednesday; that's 3 days :) [14:11:21] and approval was given on my ticket: https://rt.wikimedia.org/Ticket/Display.html?id=5474 [14:11:50] haha, um, i think 3 days == 72 hours techincally [14:11:58] its been maybe 48 :p [14:12:15] average: OpenJDK comes with jni.h as well ... for me it's for example /usr/lib/jvm/icedtea-6/include/jni.h and /usr/lib/jvm/icedtea-7/include/jni.h [14:12:35] Hehe. No fooling the ops :-/ [14:12:36] whatever :D this is just ridiculous; we just want to get qchris be abe to work [14:12:55] as if someone is going to object [14:13:09] ottomata: if you want to have a look at packages directly version 2.0.15-1 http://is.gd/1wn1Gn [14:13:54] i have jni.h [14:14:06] same error even with oracle jdk 6 [14:14:14] /usr/lib/jvm/default-java/include/jni.h [14:14:39] ottomata: can you do ./configure --enable-java [14:14:55] configure: WARNING: unrecognized options: --enable-java [14:14:59] And you have it in the correct architecture (i386 vs. x64) ... I ran into problems with that around dclass once? [14:15:06] i'm x64 [14:15:15] me too [14:15:42] qchris: I can build for i386 also [14:15:48] ok :-) [14:16:11] if I add this works: [14:16:13] -I/usr/lib/jvm/default-java/include [14:16:26] I should change the Makefile.am to include that [14:16:34] ottomata: do you get the dclass.jar too ? [14:16:46] can you try java -jar dclass.jar ? [14:16:52] well, i had to run the ./libtool command manually [14:16:54] to try it [14:17:01] so I don't htink that made the jar [14:17:06] that was not in make to get that to work [14:17:15] um, average, i'm not sure if that path is valid on other distros [14:17:17] dunno though [14:17:20] that was just a check [14:17:24] oh [14:17:27] i betcha I know [14:17:30] my JAVA_HOME isn't set [14:17:33] would that do it? [14:18:06] try setting JAVA_HOME yes [14:18:22] I should try to build on a clean VM to know what I'm missing [14:18:31] good idea [14:21:10] yes, average [14:21:14] when I set JAVA_HOME it works [14:21:17] i get dclass.jar [14:22:39] cool [14:22:40] maybe the makefile can check that JAVA_HOME has been set? [14:23:12] drdee: yes, and abort if it's not set [14:23:17] yeah that would do [14:23:28] on ubuntu you can set it to /usr/lib/jvm/default-java [14:23:30] not sure about other distros [14:24:04] ottomata: java -jar dclass.jar shouldn't work since the .so is not in place yet [14:24:26] ottomata: please pick up the 2.0.15-1 debs from here http://is.gd/1wn1Gn [14:24:38] if you install those debs, you should be able to run java -jar dclass.jar [14:24:51] install which ones? [14:25:09] ottomata: all the 2.0.15-1 debs [14:25:37] http://garage-coding.com/releases/libdclass-dev/libdclass-data_2.0.15-1_all.deb [14:25:47] http://garage-coding.com/releases/libdclass-dev/libdclass-java_2.0.15-1_amd64.deb [14:25:53] http://garage-coding.com/releases/libdclass-dev/libdclass-jni_2.0.15-1_amd64.deb [14:25:56] these? [14:25:57] libdclass0_2.0.15-1_amd64.deb [14:25:57] libdclass-data_2.0.15-1_all.deb [14:25:57] libdclass-java_2.0.15-1_amd64.deb [14:25:57] libdclass-jni_2.0.15-1_amd64.deb [14:26:00] http://garage-coding.com/releases/libdclass-dev/libdclass0_2.0.15-1_amd64.deb [14:26:06] yes [14:26:08] k [14:26:31] done [14:26:31] now? [14:26:35] java -jar dclass.jar [14:26:36] java -jar ./dclass.jar? [14:26:38] yes [14:26:46] [S] => dClassLoader.load() [14:26:46] Exception in thread "main" java.lang.UnsatisfiedLinkError: dclass.dClass.statversion()Ljava/lang/String; [14:29:00] ottomata: ln -s /usr/lib/x86_64-linux-gnu/jni/libdclassjni.so.0.0.0 /usr/local/lib/ [14:29:56] (I'll definitely have to look on this to install symlinks in /usr/local/lib/ too ) [14:30:21] same problem [14:30:58] ok I'll do it on a clean VM and come back with more details [14:31:02] thank you [14:31:09] k [14:31:10] yup! [15:19:44] New patchset: Milimetric; "this should be a functioning cohort upload" [analytics/wikimetrics] (master) - https://gerrit.wikimedia.org/r/74163 [15:19:59] Change merged: Milimetric; [analytics/wikimetrics] (master) - https://gerrit.wikimedia.org/r/74163 [15:29:12] New patchset: Milimetric; "redirect fix" [analytics/wikimetrics] (master) - https://gerrit.wikimedia.org/r/74165 [15:29:21] Change merged: Milimetric; [analytics/wikimetrics] (master) - https://gerrit.wikimedia.org/r/74165 [15:29:45] cool, this should be a good point for the demo [15:30:06] i was able to go from creating a cohort to running a job [15:30:29] if erosen can get the job results displaying, we're ballers [15:30:57] i think we can do it [15:31:18] just looking into making a url bound via knockout [15:31:25] you might have some ideas on this [16:11:02] qchris, wanna continue thinking about canary events? [16:11:35] Sure, gimme some minutes to boot the Google machine. [16:13:36] :) [16:14:57] drdee: I am in the hangout. [16:15:23] aight [16:21:23] of course, what's the day without an epic linux fail [16:21:33] hey erosen, I'm back to knockout your knockout problems [16:27:56] hehe [16:51:55] New patchset: Milimetric; "good candidate for demo" [analytics/wikimetrics] (master) - https://gerrit.wikimedia.org/r/74172 [16:52:18] Change merged: Milimetric; [analytics/wikimetrics] (master) - https://gerrit.wikimedia.org/r/74172 [17:00:49] ottomata, average: scrum [17:35:24] New patchset: Milimetric; "small jobs usability fix" [analytics/wikimetrics] (master) - https://gerrit.wikimedia.org/r/74177 [17:35:34] Change merged: Milimetric; [analytics/wikimetrics] (master) - https://gerrit.wikimedia.org/r/74177 [17:36:20] I'm in a vee em [18:11:37] qchris: i fine-tuned slide 5/6; please have a look [18:11:58] drdee: Might take some time. There are currently some gerrit problems :-( [18:12:06] okidoki [18:24:34] https://github.com/mozilla/jydoop/ [19:02:33] drdee: I just read through it. Basically looks good to me. [19:02:54] drdee: So what about I start to implement the basic components? [19:03:19] hangout? [19:03:36] drdee: I would have bet that'll be the answer :-) [19:03:43] I am already there ;-) [19:03:47] give me a sce [19:04:37] https://mingle.corp.wikimedia.org/projects/analytics/cards/806 [19:06:15] cool. [19:42:31] erosen: you around? [19:52:51] New patchset: Milimetric; "disabling non-ubiquitous revision table columns and a minor ui fix" [analytics/wikimetrics] (master) - https://gerrit.wikimedia.org/r/74257 [19:53:06] Change merged: Milimetric; [analytics/wikimetrics] (master) - https://gerrit.wikimedia.org/r/74257 [19:57:04] milimetric: hey [20:00:10] qchris: https://mingle.corp.wikimedia.org/projects/analytics/cards/807 [20:11:10] qchris: https://mingle.corp.wikimedia.org/projects/analytics/cards/808 [20:16:55] New patchset: Milimetric; "default project is required" [analytics/wikimetrics] (master) - https://gerrit.wikimedia.org/r/74261 [20:17:09] Change merged: Milimetric; [analytics/wikimetrics] (master) - https://gerrit.wikimedia.org/r/74261 [20:32:46] ERGGHH [20:32:48] i need brainbouce [20:32:50] ne1? [20:32:51] ne1? [20:32:59] brainbounce with anyone ? [20:33:03] like really anyone ? [20:33:12] ja maybe! [20:33:19] i just need to talk to someone I think [20:33:26] i got a user file permission problem [20:33:49] i am in batcave [20:33:52] or ne1 [20:33:58] for ne1 [20:36:55] not in batcave atm [20:53:26] hey, can someone remind me where the log files with user agents are on stat1? [20:53:34] (for mobile) [20:57:08] /a/squid/archive/mobile/ [21:00:21] drdee, thanks, I remember Ori mentioning that somethings wrong with the most recent files [21:00:35] is it fixed now? I guess not, July files are much smaller than June [21:00:35] not aware [21:00:45] what was the issue [21:01:15] I guess not everything is logged, or something with sampling. nevermind, I'll ask him again when I see him [21:03:25] ahh poop [21:03:27] no jgonera [21:03:29] that is still an issue [21:03:33] the mobile varnish hosts are hardcoded [21:03:37] in the filter file [21:03:42] so when mark changed them, we stopped getting proper logs [21:04:03] i fixed this for analytics log boxes, but not the older udp2log boxes [21:04:07] so that happened when? June 27? [21:04:55] yeah around then [21:06:05] fixing... [21:06:37] thanks, I'll just use mid-June data for my experiments now [21:14:01] ok, sorry about that jgonera, new data coming in from now [21:14:13] the hosts are now grabbed directly from the varnish puppet configs too [21:14:19] so if mark changes them again they should change here too [21:14:20] ottomata, no worries, thanks for fixing that [21:14:24] great! [21:14:36] laaaters! [21:15:44] thanks ottomata [21:19:05] New patchset: Milimetric; "changing Job to Report in the UI" [analytics/wikimetrics] (master) - https://gerrit.wikimedia.org/r/74269 [21:19:16] Change merged: Milimetric; [analytics/wikimetrics] (master) - https://gerrit.wikimedia.org/r/74269 [21:43:00] drdee: have you had any issues sshing into labs instances lately? [21:43:16] a debhelper is stripping one of the shared objects naked [21:43:20] naked of symbols [21:43:22] for some reason [21:43:27] I suspect dh_strip [21:45:51] erosen: let me check [21:46:33] i can login to limn0 without problems [21:46:39] (that machine needs a reboot btw) [21:48:24] average: did you see paranoid's review? [21:48:47] hrmm [21:52:39] drdee: haha [21:52:44] drdee: yes [21:52:54] drdee: you meant paravoid [21:53:16] :) [21:53:24] stupid autocorrect [21:54:03] :D [21:54:21] it's coming along, deployed 7 versions on a VM so far [21:54:38] having some problems with some .so's which are stripped naked of symbols and I have to avoid that somehow [21:57:42] ask bblack [21:57:51] asking [21:57:55] k [22:01:14] milimetric: you back? [22:01:23] wondering if you can do an ssh -vvv wikimetrics [22:01:29] and should me the output [22:01:57] drdee: when you get a chance, can you give that a try ^^^ [22:03:12] erosen: worksforme; [22:03:18] yeah [22:03:20] but can you send me the output [22:03:22] k [22:03:29] I'm trying to figure out where my setup fails [22:03:34] ty [22:04:47] sent email [22:05:06] thanks [22:06:11] drdee: is your key dsa? [22:06:44] no, rsa [22:06:49] interesting [22:07:10] k, i guess i'll ask someone in labs [22:40:04] drdee, about OpenDDR, are we using it somewhere already? is there any sample stats generated using this library? [22:40:57] we are using it through dclass, https://github.com/TheWeatherChannel/dClass [22:41:29] but we will start using the Apache Devicemap project soonish [22:41:49] dClass has been ported to Java for the Apache Devicemap proejct [22:44:42] so devicemap will replace openddr but we won't care because we use dclass as a wrapper? [22:47:00] device map will, if everything goes according to plan, unify openddr and dclass and offer bindings in multiple languages [22:47:04] but for now dclass is good [22:54:29] anyway, do we have any sample browser stats generated using dclass? in any format, limn, json, yaml, csv [22:55:07] what kind of stats? [23:01:02] well some kind of browser name/version stats (e.g. Chrome, total: X, 27.0: Y, 26.0: Z; Firefox, total: W...), from any period of time (e.g. a week from the past data) [23:01:33] btw do you know what character encoding those files in /a/squid/archive/mobile use?