[00:06:52] mk [00:36:33] dschoon: could we catch up tomorrow? I need to debrief from a few meetings today before meeting. what time are you coming in? [00:36:46] sounds good. [00:36:58] kraigparkinson: i'll be in a little before 10 [00:36:58] thanks for the flexibility [00:37:32] cool [00:37:33] no worries. [01:43:08] ciao [14:18:13] guuuuuuud meuuuuuuning [14:19:37] ottomata!!!!!!!!!! [14:19:43] moriing! [14:19:47] whoomp there he is! [14:19:53] whooomp there he is! [14:20:21] i am pumping some beats this morning [14:21:03] my body is now utterly confused about what timezone it is in. Eugh. [14:21:04] hello drdee [14:22:26] morning guys [14:22:54] mooorning [14:23:18] so ottomata, before we get that puppet stuff approved by ops, can we try it on a labs instance? [14:23:25] yesssuhhhhh [14:23:26] hey SleepyPanda, [14:23:28] it could be our kripke migration [14:23:31] i can try to make that happen [14:23:33] MILIMETRIC!!!!!! [14:23:36] i tried for a minute on friday but got annoyed [14:23:43] what's up Dr dee? [14:23:46] oh ok [14:23:54] i think drdee has had his coffee [14:23:56] that is all that is up [14:24:01] i just feel like shouting :D [14:24:10] ah yes [14:24:15] and this is without coffee [14:24:17] I don't understand the coffee people [14:24:26] mornings are for somber reflection [14:24:27] :) [14:24:29] aha [14:24:40] just took my first sip of coffee [14:24:43] coffee is just yummy [14:24:53] food wakes me up more than coffee [14:25:06] it's more the ritual than anything else (IMHO) [14:25:24] yeah, i like eat sugar wrapped in as much fiber as necessary to prevent it from killing me :) [14:25:52] sooooo ottomata, how eager are you to deploy the filter component of webstatscollector with the replace function? We actually received an official bugzilla bug about this :) [14:26:02] so is there anything I can try with labs? do you just spin up an instance and run the puppet stuff on it manually? [14:26:12] oh no - you said there was some puppet master [14:26:14] drdee, suuure we can do that [14:26:23] we just revert if things break [14:26:28] milimetric [14:26:29] https://labsconsole.wikimedia.org/wiki/Help:Self-hosted_puppetmaster [14:26:31] but then at least we have tried :) [14:26:38] cool ottomata, I'll take a look [14:26:54] i have an instance with most of those instructions done already [14:27:08] puppet-dev.pmtpa.wmflabs [14:27:20] milimetric: about the limn sprint, very few tasks seem to have been finished so far, is that correct? [14:27:40] that's right [14:27:48] it's all puppet/deb that's in process [14:27:58] but i feel like a lot got done [14:28:18] how far are you guys with the debianization? [14:28:20] i will add a few tasks that got done in order to make that happen [14:28:24] ok [14:28:25] deb is done [14:28:31] nice! [14:28:36] without npm? [14:28:37] we just wanted to wait to test on puppet [14:28:39] without yea [14:28:42] k [14:28:57] did average_drifter do the debianization? [14:31:05] milimetric [14:31:22] i'm going to try puppet dev stuff now, i was just talking with hashar about the bit i wasn't sure how to do [14:33:18] cool [14:35:52] average_drifter: can you review https://gerrit.wikimedia.org/r/#/c/47827/ [14:47:46] agh, milimetric, i think i borked this instance when I was playing with it friday, i'm going to trash it and start a new one, unless you are interested in trying to follow those instructions…? [14:50:12] um, sorry ottomata - no I'm looking at something else atm [14:50:18] destroy away [14:51:13] mk [15:06:24] ottomata: wondering if you can help me with some issues building the kraken? [15:06:33] i would ask drdee, but he doesn't seem to be around [15:07:03] ottomata: have you successfully compile the kraken jars lately? [15:08:17] no [15:08:23] but i haven't tried very much [15:08:24] actually [15:08:24] yes [15:08:26] i got pretty clos [15:08:30] but I had to skip tests [15:08:31] hehe [15:08:34] mvn package -DskipTests [15:08:35] i think [15:08:39] yeah that ws my iussue [15:08:40] yay [15:08:44] i'll give that a try [15:08:46] k [15:08:58] or do you know a way to compile only specific parts? [15:09:41] actually nvm, i thiknk running mvn compile in the subproject directories worked cause they have their own poms [15:10:39] aye [15:22:52] ottomata: another kraken / maven q: do i need to do anything special to use the wmf nexus instance? [15:23:03] not that I know of [15:23:06] k [15:23:13] i'm getting this error: Could not resolve dependencies for project org.wikimedia.analytics.kraken:kraken-pig:jar:0.0.1-SNAPSHOT [15:23:32] and i suspected it had to do with not having the the right repositories available [15:23:39] yeah, i had that problem tooooooooo, but I thought drdee fixed it for me yesterday [15:23:48] interesting [15:24:45] i think i need to put a settings.xml file somewhere [15:24:53] i found this one in kraken: https://github.com/wikimedia/kraken/blob/master/maven/example.settings.xml [15:27:27] oo that sounds familar, maybe somewhere in ~/.m2 [15:27:28] hmm [15:27:35] yeah [15:27:38] put that in there [15:27:39] i think [15:27:59] trying now [15:31:51] guys, analytics1027 is being really weird with cron, and I'm suspcious of the recent ldap nss removal stufff [15:32:02] i'm going to reboot it, but this means oozie is going down for a min [15:32:49] ottomata: succes! settings.xml in ~/.m2 did it [15:35:50] yeehaw [15:40:31] drdee, re: that sql query for active editor skin usage [15:40:56] I feel like we should have an up to date SQL slave, whether or not we do anything with it [15:41:19] it would be useful to see what the rest of the org sees, and that seems to be how most people start thinking about analysis [15:41:28] so, what can I do to help make that happen? [15:42:33] milimetric: drdee isn't around afaik, but if you want access to the db slave, you could probably get access to the s*-analytics-salves [15:42:57] is that something that everyone else has? [15:43:04] not everyone [15:43:08] i mean our team [15:43:11] the "analysts" do, though [15:43:26] ok, yeah that sounds good. Any idea who to ask? [15:43:27] i suspect diederik has access, but otherwise, i might be the only one [15:45:38] i think Dario is a good contact, let me find the relevant wiki page [15:45:39] i'll ask Dario, no worries [15:45:42] can't find it. I think they are just credentials--not user specific accounts [15:45:42] k [15:45:42] cool [15:45:42] so in theory I could share mine with you, but I should check the norms [15:45:53] yeah, i'll go through Dario to make sure we're on the up and up [15:45:54] :) [15:57:32] dschoon: you know anything about the kraken build process? [16:08:06] milimetric, can you show me how to set up dev-reportcard with limn pointing at a tsv and pivoting? [16:08:11] http://stats.wikimedia.org/kraken-public/webrequest_loss_by_hour.tsv [16:08:49] the pivot happens automatically when the datasource refers to a datafile and defines format: 'log' [16:08:57] but I can help [16:09:00] hangout? [16:09:07] standup link [16:09:42] k [16:37:52] HMM [16:37:53] milimetric [16:38:00] after installing the .deb (and apt-get isntal [16:38:00] so [16:38:03] I did [16:38:14] apt-get install nodejs rlwrap [16:38:20] sudo dpkg -i limn...deb [16:38:26] but [16:38:42] /usr/bin/env: coco: No such file or directory [16:38:45] average_drifter ^ [16:43:12] ottomata: looking at it [16:43:30] ottomata: can you link me to the deb you used please ? [16:47:09] uh, its the one milimetric just emailed me [16:47:12] i can email it to you [16:47:35] oops, was downstairs [16:47:48] also, you should be able to ssh into puppet1.pmtpa.wmflabs [16:47:50] where i'm working on this [16:47:53] cool [16:47:55] the .deb is in my homedir [16:47:55] doing so now [16:50:02] trying the dpkg -i myself now [16:51:11] ottomata: I know the reason [16:51:22] the reason is this [16:51:22] oh ja? [16:51:34] me and Dan talked about coco [16:51:41] he said "do npm install -g coco" [16:51:56] and I said "hey, but let's have it self contained within the node_modules" [16:52:02] I tried to have it in node_modules [16:52:15] oh I see the confusion [16:52:17] but then coco was complaining that it couldn't find server.js [16:52:24] I was saying install -g coco was ok on the Build box [16:52:27] so because we had to pull out a deb I decided to -g coco [16:52:32] but no coco sits in /usr/lib somewhere [16:52:38] and we haven't included it in the .deb [16:52:45] i think it is in the .deb [16:52:53] ottomata: let me check [16:52:54] just not in the path [16:52:57] it should be included by default in node_modules [16:53:10] moment, I'll check [16:53:23] /usr/lib/limn# ls -l node_modules/.bin/coco [16:53:23] lrwxrwxrwx 1 root root 22 Feb 20 16:18 node_modules/.bin/coco -> ../coco/lib/command.js [16:53:39] server.co does [16:53:43] #!/usr/bin/env coco [16:53:54] but node_modules/.bin/coco is not on the path [16:54:05] need wrapper script :p [16:54:12] wrapper script ? [16:54:31] ottomata: oh btw, remember you gave me a gist with the limn.conf ? [16:54:36] yes [16:54:39] ottomata: that one is in debian/limn.upstart now [16:54:46] ok [16:54:49] ottomata: did you build from develop branch ? [16:54:59] I thought we were just going to do $LIMN_INSTALL/node_modules/.bin/coco server.co to start it [16:55:08] milimetric: no, because it doesn't work [16:55:09] oh we were? [16:55:13] no [16:55:20] why doesn't that work? [16:55:24] because coco doesn't like that [16:55:25] works locally for me [16:55:34] it works i think too [16:55:37] milimetric: it works locally for you because you already have coco installled [16:55:37] i just tried [16:56:01] ottomata: did you do npm install -g coco on the machine where you tried that ? [16:56:12] because if so, then that means coco is able to find its stuff in /usr/lib... [16:56:22] mmm, no, I just npm remove -g coco and it still works [16:56:41] /usr/lib/limn/node_modules/.bin/coco /usr/lib/limn/server/server.co [16:56:42] path.existsSync is now called `fs.existsSync`. [16:56:42] Limn Server (port=8081, env=development, rev=8780a4f, vardir=./var, data=./var/data) [16:56:44] no i did not [16:56:45] ok, time to kill the coco myths [16:57:07] node_modules/.bin/coco is the same exact thing as the global coco you get when installing with npm i -g coco [16:57:26] milimetric: try removing all the cocos you have on your machine [16:57:29] if it's not working in some environment, let me know where and we'll figure out what the problem is [16:57:35] milimetric: and then try the node_modules one [16:57:38] I did average_drifter, they're all gone [16:57:40] milimetric: if that works then I'm wrong [16:58:00] I removed it and now if I go "coke server" it doesn't know what coke is [16:58:01] i never installed coco on puppet1 [16:58:03] and that worked [16:58:13] but node_modules/.bin/coco still works [16:58:18] milimetric: doesn't coco pull coke too ? [16:58:34] I remember dschoon told me coco pulls coke too [16:58:40] pull? [16:58:42] as a dep [16:58:55] npm i -g coco installs coke, yes [16:59:04] but coke isn't needed for running the server, just building and stuff like that [16:59:35] ok, I'll build from develop and install the deb [16:59:42] oh crap [16:59:47] milimetric: ? [16:59:49] but without coke we will have to manually link the data [16:59:55] because coke link_data does that for us [17:00:05] milimetric: so we need coke too right ? [17:00:16] inside the .deb [17:00:19] but that's not something we can help right? Ops doesn't let us do npm i -g coco right ottomata? [17:00:31] I don't think we can put coke inside the deb unfortunately [17:00:54] milimetric: but do we really need coke ? [17:01:10] no, it's not strictly necessary, it's just nice [17:01:19] ok [17:01:28] I'll try to build the deb and see if I get the same error [17:01:28] right [17:01:34] also [17:01:37] average_drifter [17:01:38] I'll let you know in a couple of minuts [17:01:41] when I remove the .deb [17:01:41] ottomata: yes [17:01:43] uninstall [17:01:45] yes [17:01:51] $ sudo dpkg --purge limn [17:01:51] (Reading database ... 38788 files and directories currently installed.) [17:01:51] Removing limn ... [17:01:51] Removing user `limn' ... [17:01:51] Warning: group `limn' has no more members. [17:01:51] Done. [17:01:51] Purging configuration files for limn ... [17:01:52] /usr/sbin/deluser: The user `limn' does not exist. [17:01:52] dpkg: error processing limn (--purge): [17:01:53] subprocess installed post-removal script returned error exit status 2 [17:01:53] Processing triggers for ureadahead ... [17:01:54] Errors were encountered while processing: [17:02:16] ottomata: I'll fix that too [17:02:22] ok cool [17:02:39] it's the deluser inside the postrm script [17:02:49] yeah it look slike it tries to remove it twice [17:03:08] maybe the deb is smart and it already removes it and then the postrm tries to do that again [17:03:09] so yeah, we should make the upstart script [17:03:18] start by calling node_modules/.bin/coco [17:03:18] ottomata: the upstart script is already made [17:03:25] instead of just server.co [17:03:29] right, but that will fix the coco problem [17:03:35] ottomata: what branch are you building the .deb from ? develop right ? [17:03:44] i htink so, whatever milimetric gave me [17:03:45] i'm not building it [17:03:55] milimetric: what branch have you built the .deb from ? [17:03:56] one sec average_drifter [17:03:57] I'll push [17:04:00] ok [17:04:16] ok, pushed [17:04:20] it's develop yes [17:04:27] just had one commit not on /wikimedia [17:04:33] yeah, average_drifter [17:04:38] this in the upstart inti works [17:04:38] exec ./node_modules/.bin/coco ./server/server.co [17:05:43] hi all [17:06:16] kraigparkinson: hello Kraig [17:08:00] milimetric: sorry I'm late. [17:08:15] no prob kraigparkinson, I'm on the standup hangout [17:09:10] if it doesn't work, you can start a meeting and invite me instead [17:20:12] ok guys, i'm moving to a cafe [17:20:23] fyi, aside from that coco startup issue, the .deb and puppetization seem to work! [17:21:24] ottomata: cool, we'll get the coco startup sorted out in the next hour or so [17:40:41] yo drdee: where did you end up putting that kraken jar yesterday? [17:40:51] no where :) [17:41:07] hehe, i suspected [17:41:16] i attempted to build it myself and ran into some issues [17:41:20] have a sec? [18:18:56] * brion waves hello [18:19:13] zomg a brion :D [18:19:17] I'm looking for some usage figures on iOS browser versions [18:19:27] http://stats.wikimedia.org/wikimedia/squids/SquidReportOperatingSystems.htm has comparative versions but was last generated in november [18:19:33] is there anything more up to date i can grab? [18:20:08] ideally, i want to decide that iOS 5 usage is low enough i don't have to support it for new apps ;) [18:20:16] brion: very alpha [18:20:17] http://test-reportcard.wmflabs.org/graphs/pageviews_mobile_by_manufacturer [18:20:20] love feedback [18:20:28] we are about to send this to the mobile team anyways [18:20:37] browser version are very high on the priority list [18:20:40] pretty! [18:20:48] data view is broken :( [18:20:53] but we love love feedback [18:20:57] data is 15 minutes old [18:21:19] yeah lots of NaN :) [18:21:29] we know :) [18:21:38] you can download the ison file [18:22:26] nice, i like being able to see different breakdowns by country [18:22:43] this is still vendor / manufacturer [18:22:45] that would be more useful to me on an OS and OS version basis than manufacturer [18:22:53] *nod* [18:22:57] drdee: Apple is... 20 times bigger than Samsung? [18:23:03] yes, we are on it [18:23:29] YuviPanda: we might need to tweak the device recognizer more [18:23:45] YuviPanda: rumor is apple users surf the web more than android users on average. don't know how true this is, that's a big difference indeed [18:23:51] hmm, and Samsung is only slightly bigger than apple in India [18:24:02] which definitely doesn't sound too right [18:24:12] brion: true, but 20 times? [18:24:28] yeah :) [18:24:31] remember: this is alpha and this is exactly the feedback we need [18:24:41] yeah, so giving :) [18:24:49] drdee: also I can't see the full list - my screen is too small, I guess? [18:24:51] (13") [18:25:08] average_drifter, where can I find lib-dclass .deb? [18:25:12] that could be a limn bug [18:25:13] gets cut off after Samsung on India. I don't see the end of the popup. [18:25:15] or how can I get it on my mac? [18:25:17] milimetric ^^ [18:25:37] right there is 'desktop' which I'm not seeing. [18:25:44] I can see it if I zoom all the way out [18:25:55] oh more (I found motorola after desktop) [18:26:18] drdee: and 'see a table' is very Batman-ish [18:26:25] 0NaN-NaN-NaN NaN:NaN [18:26:34] yes that's broken [18:27:06] ottomata: https://github.com/wikimedia/dClass/tree/package [18:27:15] yeah and clips too. like I can't see any numbers for australia [18:27:36] can I maybe jsut clone and make && make install? [18:27:40] average_drifter: can you review https://gerrit.wikimedia.org/r/#/c/47827/ [18:28:01] ooh this is an awesome version marketshare graph… http://thetechblock.com/wp-content/uploads/2012/11/share-count-10.jpg [18:28:13] drdee: yes sir, I'll have a look at it [18:29:28] hmm, doesn't look like dClass will compile on os x [18:29:36] ottomata: oh yes it will [18:29:41] oh? [18:29:45] ottomata: drdee was able to compile it on OSX [18:29:45] main.c:73: error: ‘CLOCK_REALTIME’ undeclared (first use in this function) [18:29:54] #I also compiled it on [18:29:57] OSX [18:30:25] that problem was long ago fixed [18:31:05] ottomata: check_1) have you switched to package branch ? [18:31:24] ottomata: check_2) look at https://github.com/wikimedia/dClass/blob/package/Makefile.am , it does not involve compiling main.c [18:31:39] ah i did not switch branches, sorry [18:31:47] ok [18:33:07] dschoon, when I am in SF this/next week [18:33:15] can you help me get setup with a good java env [18:33:20] every time I try to do anytihng java I just give up [18:33:37] just install intellij idea [18:33:39] all I want to do is edit java code and compile, now i'm down the rabbit hole again [18:34:08] i've got intellij [18:34:14] not using it though cause I don't know how :p [18:34:21] average_drifter [18:34:22] ok [18:34:24] i'm in pacakge branch [18:34:26] what do I do? [18:34:34] there's no info about compiling [18:34:36] just info on how to build .deb [18:34:41] which I obvi can't do on os x [18:34:55] ottomata: ok, do this [18:35:44] ottomata: libtoolize; [18:35:44] aclocal; [18:35:44] autoheader; [18:35:44] autoconf; [18:35:44] automake --add-missing; [18:35:54] make [18:36:01] now look in .libs/ [18:36:06] #you should find the .so you need there [18:36:13] now make install [18:36:19] -bash: libtoolize: command not found [18:36:20] :p [18:36:22] brew? [18:36:25] ottomata: brew [18:36:34] brew or port [18:36:40] k [18:36:44] whichever you prefer [18:38:34] make: *** No targets specified and no makefile found. Stop. [18:38:49] in src? [18:38:58] configure? [18:39:05] configuring [18:39:22] close [18:39:23] jni/dclass-wrapper.c:4:17: error: jni.h: No such file or directory [18:39:48] yes [18:39:50] moment [18:40:05] ottomata: if you look at Makefile.am [18:40:10] there's a JAVA_HOME there [18:40:15] you need to set that in your .bashrc [18:40:20] you probably don't have that variable [18:40:38] check_3) make sure $(JAVA_HOME)/include/linux exists [18:40:50] check_4) make sure $(JAVA_HOME)/include exists [18:40:52]