[07:50:52] New patchset: Ori.livneh; "Experiment: set defaultremote=origin in .gitreview" [analytics/glass] (master) - https://gerrit.wikimedia.org/r/67240 [07:51:45] Change merged: Ori.livneh; [analytics/glass] (master) - https://gerrit.wikimedia.org/r/67240 [10:34:53] !notify ottomata [12:41:56] mooooorning ottoamta [12:48:15] mornign! [12:59:02] iiiIIinteresting, ok drdee [12:59:09] so for the last to nights [12:59:30] …..packetloss? [12:59:36] there has abeen a single hour of negative packetloss [12:59:36] yeah [12:59:38] but negative [12:59:43] which previously meant duplicates [12:59:48] yeah i was gonna ask you about that [12:59:50] but i checked yesterday [12:59:53] i saw it [12:59:55] and didn't find duplicates, at least for one host [13:00:02] and today it lasted for two hours [13:00:23] just an hour, no? [13:01:50] yeah only for an hour [13:02:20] negative loss just means that there were more lines counted in an hour than the max seq - min seq [13:02:22] for that hour [13:02:52] yesterday I examined the logs for the hour for duplicates, and didn't find any [13:04:06] mmmmmmm [13:04:36] yeah you are right, it was for 1 hour (58 minutes to be exact) [14:00:19] hey average, are you around? [14:27:08] ottomata: looking at http://ganglia.wikimedia.org/latest/graph_all_periods.php?c=Analytics%20cluster%20eqiad&h=analytics1003.eqiad.wmnet&r=hour&z=default&jr=&js=&st=1369741866&v=-99.070396&m=webrequest_loss_average&vl=%25&ti=Average%20Loss%20Percentage&z=large it says "No matching metrics detected" [14:28:19] that is a different metric than the one that just went stale [14:28:36] that only exists on analytics1027 [14:28:40] that's the one that inspects hdfs [14:28:52] [travis-ci] develop/35e70e2 (#138 by milimetric): The build has errored. http://travis-ci.org/wikimedia/limn/builds/7842500 [14:29:44] ok, sorry [14:30:17] ja dunno what's happening [14:32:09] as far as I can tell its fine [14:32:09] http://ganglia.wikimedia.org/latest/graph.php?r=hour&z=xlarge&title=&vl=&x=&n=&hreg[]=analytics100%5B3456%5D&mreg[]=packet_loss_average>ype=line&glegend=show&aggregate=1&embed=1&_=1370529080896 [14:32:18] its just icinga that can't get the value from ganglia [14:32:22] but that seems weird [14:37:01] ah yeah [14:37:02] drdee [14:37:07] gadolinium is STALE too [14:37:07] hm [14:37:41] mmm [14:46:11] yeah, drdee, it looks like all of the icinga ganglia alerts are stale [14:46:20] k [16:06:44] hiaa average [16:06:45] you around? [16:09:42] ottomata: hey [16:09:43] i'm here [16:11:40] hiyaaa [16:11:42] q [16:12:02] does your dclass makefile compile dclass? or just the jni stuff? [16:12:20] after running make I only see sos for jni files [16:13:54] ottomata: it compiles the needed objects and then links them into the .so [16:14:00] the .so should be located in .libs/ [16:14:10] i see .sos there with jni in the name [16:14:15] shouldn't there be non jni .sos? [16:14:17] just for C? [16:15:23] ohh, i need to run make in the src dir [16:15:24] hm. [16:15:34] should your global Makefile do that too? [16:15:35] also [16:15:37] why .libs? [16:15:42] also [16:15:45] does make install work? [16:15:47] what does it do? [16:15:49] (I haven't tried) [16:17:14] hehe, sorry, too many questions at once [16:17:18] wanna hangout? [16:22:04] sure [16:24:51] milimetric: want to chat about the umapi upload bug? [16:33:53] (sorry, average, now talking with alex about kafka .deb [16:34:17] ottomata: ok [16:39:57] New review: Erosen; "some comments on the unicode stuff, feel free to disregard, as they are mostly aesthetic." [analytics/user-metrics] (master) - https://gerrit.wikimedia.org/r/67136 [16:42:38] hey erosen, just got some food [16:42:49] oh that bug was simple [16:43:09] well, my partial fix was at least [16:43:44] so I am now stripping any whitespace plus \xc2 and \xa0 off the usernames, as well as uppercasing the first letter [16:43:57] average: https://plus.google.com/hangouts/_/a82d5348152766ee29bf53739023abe74b73b14b?authuser=1 [16:44:04] i'm still kinda chatting with alex [16:44:04] but i'm there [16:58:46] oh man! bitten in the ass by double letters! :) [16:58:58] git commit -ammend will commit all with message "mend" [16:58:59] lol [17:00:50] New patchset: Milimetric; "fixed upload capitalization and whitespace stripping" [analytics/user-metrics] (master) - https://gerrit.wikimedia.org/r/67136 [17:01:16] milimetric, erosen, average: scrum [17:01:39] joining now [17:05:13] drdee: how do I followup on that wmfwiki underreporting thread? do you need more evidence and how to find it? should a bug be filed? do you think it's all ok? [17:05:41] yeah it looks weird, please file a bug [17:13:15] New review: Erosen; "looks good. sorry about the .title() oversight; good catch, dan." [analytics/user-metrics] (master); V: 2 C: 2; - https://gerrit.wikimedia.org/r/67136 [17:13:16] Change merged: Erosen; [analytics/user-metrics] (master) - https://gerrit.wikimedia.org/r/67136 [17:14:13] filed as https://bugzilla.wikimedia.org/show_bug.cgi?id=49266 , for the channel records [17:17:49] milimetric: was that appropriate for me to merge the change? or should I leave that to you? [17:59:15] average, lemme know when you want to chat about dclass [18:09:43] oh totally cool that you merged it erosen [18:12:58] milimetric: i'm going to look into creating a test labs instance. does that sounds good? [18:13:28] sure, then after the metrics meeting we can get back together on the db/testing/queue [18:14:20] sounds good [18:15:31] New patchset: Stefan.petrea; "Fixing problem with cronjob [IN PROGRESS]" [analytics/wikistats] (master) - https://gerrit.wikimedia.org/r/67010 [18:26:18] milimetric: https://wikitech.wikimedia.org/wiki/Nova_Resource:I-000007b0 [18:26:29] yay [18:26:30] nothing interesting there yet [18:26:37] course [18:26:38] but i logged-in! [18:26:42] :) [18:39:08] milimetric: is there a meeting with the topic of phasing out wikistats ? [18:39:18] milimetric: if so, what is the name of it ? [18:45:07] nothing I know of average_drifter, just a general idea that we'd like to move to code that's easier for all of us to maintain [18:52:42] average_drifter: yoyoyoyoyooyyooyoyo [18:52:43] yooooo! [18:52:44] yoweeee! [18:52:49] what's doin? [18:55:22] erosen: you're going to lunch now right? [18:55:54] let's set up a time to hangout right after? [18:57:25] yeah [18:57:26] sounds good [18:57:47] milimetric: how about 12:30? [18:57:55] works :) [18:58:20] cool [19:07:57] ottomata: talking to Adrian [19:08:22] milimetric: sent you a draft of a question to labs-l [19:08:22] cool [19:09:10] yeah, that looks good erosen [19:09:22] you can cc me [19:09:24] yeah [19:30:59] milimetric: joining hangout now [19:31:48] one sec erosen [19:32:07] no rush [19:32:13] we can reschedule [19:32:29] you're just usually already hanging out when i join [20:29:05] average, dclass? [20:29:13] average_drifter: dclass? [20:46:45] ottomata: moment [20:48:09] hokaayyy [21:14:58] the lady demandeth food erosen :) [21:15:14] i pushed the refactor to in-memory sqlite mediawiki tables [21:15:20] tests confirm it's working [21:16:31] I'll be back working in a bit, but in the meantime you could look at flask setup and/or implementing namespace_edits [21:16:36] I'll write the tests for namespace_edits [21:43:13] hey ottomata [21:43:54] hiya [21:44:17] I haven't heard back from you guys on the stat1->stat1001 rsync issue, any ETA to have this up and running? [21:44:22] howdy [21:45:58] ottomata: I hadn't had a chance to look into it fully, but things on stat1 seemed okay (directory present, right permissions, etc.) -- it's whether the rsync job was created on stat1001 that's the question, and I don't have SSH access to that box, so I can't look into it myself. [21:46:15] sooooooooooo i think you have to ask drdee, i've been told to focus on things that are planned for sprints [21:46:35] ic, I pinged him already but I'll insist ;) [21:46:42] k you can ask adrian too [21:46:48] this is becoming a blocker as the TS is totally unreliable [21:47:19] I think he's copied on all my messages, but I'll double check - thx [21:47:48] ori-l: is this the limn mobile data sync? [21:48:13] aside from the email thread, i'm not sure I know what rsync job you are referring too [21:48:32] no, this was another patch. let me take a look at the state of the puppet configs and get back to you in a couple of mins. [21:49:44] ottomata: this is misc::statistics::public_datasets, specifically the cron resource in statistics.pp line 197 [21:49:53] command => '/usr/bin/rsync -rt stat1.wikimedia.org::public-datasets/* /var/www/public-datasets/', [21:50:01] this should be running on stat1001 [22:09:16] ah, ori-l [22:09:22] @ERROR: Unknown module 'public-datasets' [22:09:34] using the :: in a rsync url expects there to be an rsync daemon module set up with that name [22:10:21] oh, right. would you like me to add an rsync module, or is it easier to replace the '::' in the url? [22:10:26] ottomata: ^ [22:10:55] i think there is already one for wwww [22:10:55] one sec [22:10:55] the module is easier because then there are no ssh issues [22:12:00] so [22:12:01] you want [22:12:16] stat1:/a/public-datasets -> sta1001:/var/www/public-datasets, right? [22:12:20] ori-l ^ [22:12:22] oui [22:12:26] k... [22:13:38] ottomata: DarTar or I can write back to drdee to make sure the time you spent diagnosing / fixing is accounted for [22:13:48] haha [22:13:49] yeah, I was going to ask [22:13:49] ok [22:14:17] think its ok, i'm kinda done working for the day and this is real easy, this isn't a new thing [22:14:17] rounce123 ^^ [22:14:19] just a quick fix [22:14:27] jesus loves you [22:15:47] and he knows you're right [22:19:52] DarTar [22:20:00] yo [22:20:06] you called [22:20:31] rounce123: DarTar and I bullied ottomata into fixing my puppet glitch [22:20:39] want to make sure he gets the requisite credit [22:20:52] He will :-) [22:21:00] thanks :) [22:21:07] haha [22:21:09] :) [22:21:39] ori-l: DarTar: http://stat1001.wikimedia.org/public-datasets/data-test [22:21:56] VICTORY [22:22:12] sweeeeeeet [22:22:37] * DarTar syncs a couple of beers to brooklyn [22:22:49] s/syncs/rsyncs [22:23:01] haha, put the -t flag on it so they stay cold and yummy! [22:24:54] ottomata, ori-l: we should expect this to run every 45 mins, right? [22:25:05] if I remember from ori-l [22:38:23] i think so ja [22:38:38] or no [22:38:38] every hour [22:38:40] every :45 [22:41:20] lataaaS! [22:45:11] New patchset: Stefan.petrea; "Fixing problem with cronjob [IN PROGRESS]" [analytics/wikistats] (master) - https://gerrit.wikimedia.org/r/67010 [23:15:38] Change merged: Stefan.petrea; [analytics/wikistats] (master) - https://gerrit.wikimedia.org/r/67010 [23:17:20] milimetric: I noticed that when I try to sign it at http://ee-dashboard.wmflabs.org/dashboards/features I get redirected to http://127.0.0.1:8088/ upon signing in, is this a known issue?