[00:19:48] New patchset: Milimetric; "a bit more testing added for job request" [analytics/wikimetrics] (master) - https://gerrit.wikimedia.org/r/71943 [00:19:59] Change merged: Milimetric; [analytics/wikimetrics] (master) - https://gerrit.wikimedia.org/r/71943 [00:21:26] nite everyone, 696 is in sign-off [00:40:25] laterz [06:46:29] New patchset: Rfaulk; "fix. pep8." [analytics/user-metrics] (master) - https://gerrit.wikimedia.org/r/71963 [06:46:57] Change merged: Rfaulk; [analytics/user-metrics] (master) - https://gerrit.wikimedia.org/r/71963 [13:16:27] oof, drdee, something is totally bad [13:16:30] i'm looking into it [13:16:35] but whatever I did last night did not fix [13:46:35] Hi guys, just to present myself, I'm currently working in Bitergia helping in the community dashboard [13:46:49] I'll try to be around to put ideas together :) [13:47:02] a pleasure to work with you [13:53:53] Hi cool! nice to meetcha [14:02:53] hey ottomata [14:02:53] he qchris [14:03:01] ottomata brain bounce? [14:16:12] ottomata? [14:16:46] :) [14:35:59] hyia [14:36:25] sure [14:36:26] drdee [14:36:40] k [16:28:36] erosen, i think I have been importing the wrong data since mark deployed new mobile varnishes [16:28:37] since june 25 [16:28:42] aye [16:28:51] interesting [16:28:59] what sort of wrong data [16:29:06] like, logs from the wrong varnishes [16:29:22] are there other varnishes? [16:29:30] i needed to import from cp1046, cp1047, cp1059, and cp1060 [16:29:31] i guess the answer is yes [16:29:41] what were you collecting instead? [16:29:48] so, i'm using a regex to match hostname [16:29:58] yea [16:30:00] and the match for the cp1046 and cp1047 bit [16:30:01] was [16:30:05] cp10[67] [16:30:07] missing the 4 [16:30:13] so [16:30:17] aa [16:30:26] that would match cp1060-cp1079 [16:30:29] but I guess I'm wondering what the traffic is on those machines? [16:30:44] i was under the impression that we were getting all the varnish traffic [16:30:48] bits [16:30:51] ? [16:30:51] aah [16:30:59] by [16:31:10] hmm that's 69 and 70 [16:31:21] "getting al the varnish traffic", I meant, I thought varnish was just for mobile [16:31:24] but clearly it is for bits too [16:32:27] naw [16:32:30] and i'm checking now [16:32:37] and i think there are acutally only those two hosts in that range [16:32:41] so I got 60 [16:33:04] and 59 [16:33:09] but not 46 and 47 [16:33:24] also got 69 and 70, which are bits varnishes [16:34:15] ottomata: i guess the essential question is whether there is a method to reconstruct estimates of the true counts, and whether it is worthwhile [16:34:30] it seems like we should be able to filter out all of the bits requests by url [16:35:03] and then use the traditional fraction of traffic served by the two hosts we were getting to find a normalizing constant [16:41:26] yeah [16:41:30] so basically [16:41:39] we captured 50% of mobile traffic during that time [16:41:40] so [16:48:16] yeah, we can filter out the non mobile hosts, crunch numbers, and multiple x 2 [16:48:20] but yuck. [16:52:45] yeah [16:52:57] it's not actually that bad [16:53:02] if we know the date range [17:01:02] erosen standup? [17:01:02] average, erosen: scrum [17:01:20] http://cl.ly/image/201B0F1s270q [17:01:23] drdee: hi, I'm in the progress of making a draft of modules needed, diagnosing the problem on stat1002, and working on the new comments in the gerrit patchset https://gerrit.wikimedia.org/r/#/c/68711/ [17:01:36] average, standup? [17:01:45] or is that your update? :p [17:01:49] this is my updated yes [17:01:52] I cannot attend today [17:02:33] will get home in 1h and then I'll be able to attend hangouts if needed [17:03:28] ok cool, no prob [17:04:14] average: you should really try to attend scrums [17:56:50] New patchset: Matthias Mullie; "Generate data for Echo preferences" [analytics/limn-ee-data] (master) - https://gerrit.wikimedia.org/r/71999 [18:18:26] drdee: regarding #789. It's a stupid question, but does "filtering for canary event" work if udp2log does sampling? [18:18:54] yes because we would inject the canary event after the sampling step [18:19:02] and we are migrating towards unsampled streams [18:19:35] if we do it after the sampling, the we would not catch problems at varsnishncsa would we? [18:19:36] ? [18:19:45] qchris, i think that's right [18:20:08] so [18:20:15] you guys wanna brainbounce this real quick? [18:20:18] i have some thoughts [18:20:23] sure give me 5 minutes [18:20:25] k [18:20:34] i haven't had breafast [18:22:34] yeah i haven't had lunch [18:22:44] I am lacking dinner :-) [20:53:00] drdee, qchris, i got a few mins if you wanna brain bounce real quick about canaries before I sign off for the day [20:53:21] i am waitiing [20:53:30] i am here as well :-) [20:54:55] ottomata: Shoot :-) [20:56:46] https://plus.google.com/hangouts/_/2da993a9acec7936399e9d78d13bf7ec0c0afdbc [20:56:59] Oh ... you're using the google thing again :-) [20:57:03] 1min.