[13:21:51] mooooorning [13:34:37] moooorning! [13:34:44] oh look so many emails! [13:34:52] love letters? [13:37:14] maybe! [13:37:15] who knows!? [13:37:32] woot woot, woot woot [14:10:57] heya, drdee, do you have analytics-l mailing list admin login? [14:11:05] i sure do [14:11:06] i keep getting emails saying that I need to approve messages [14:11:13] i probably did that already [14:11:18] most are spam, but there's one in here from thursday that looks good [14:11:19] oh ok [14:11:24] although there might be a be small backlog from this weekend [14:11:32] Page view stats crashing on https English Wikipedia articles ? [14:11:36] oh yhea [14:11:37] i see it [14:11:40] yeah i approved that one [14:11:41] it made it through [14:11:41] ok cool [14:11:54] how's life man? [14:12:01] when are you flying to AMS? [14:12:09] milimetric: when are you flying to AMS [14:12:16] it's good to have you back ottomata! [14:12:27] the 22nd, but i think i'm getting there the 23rd [14:12:29] i'll check [14:12:36] ottomata: check this out: https://gerrit.wikimedia.org/r/#/c/63220/ [14:12:48] you are going to give me a virtual hug [14:14:51] yep, i'm getting into AMS at 8:00 am on the 23rd [14:15:36] aight, so you will be jetlagged [14:15:51] and when will you leave? [14:26:13] milimetric: , awesome! I'm getting there the 22 too! [14:26:21] i'll be there around noon on the 22 [14:26:40] and i'm not sure when I'm leaving yet, probably that sunday eve [14:26:46] i'm leaving monday morning the 27th [14:26:59] cool, we should do fun stuff :) [14:28:37] yes we should! [14:28:48] i've contacted the polo people in amsterdam, seems there are so many of them there [14:28:52] but there will be at least one polo night! [14:30:42] how exhausted will you be on the 22nd? [14:32:57] dunno! [14:33:06] probably purty sleepy, i don't usually sleep well on planes [14:33:32] but I also like to fight jetlag by powering through ti [14:33:33] it [14:33:36] we will see! [14:33:44] ok, drdee, emails read. [14:33:55] should I continue to work on stat1002? [14:33:57] today? [14:34:39] sooooo....... [14:35:15] do you need to chime in on the kafka convo with faidon? [14:35:28] i did some research this weekend [14:35:50] then a review of https://gerrit.wikimedia.org/r/#/c/63220/ would be great as well [14:35:53] yeah I read, sounds good, but 0.8 isn't stable, is it? [14:36:00] how much work is left for stat1002? [14:36:31] about kafka, that's right but we are going to use 0.8 anyways; that's what Magnus is working on and according to him it should be released any omen now [14:36:36] omen -> moment [14:37:07] after https://gerrit.wikimedia.org/r/#/c/63220/ has been merged we should re-enable udp2log for the nginx servers [14:37:32] we are effectively no longer counting ip6 and ssl based traffic in webstatscollector [14:37:38] for stat1002, not sure, i need to make sure the stuff copies over properly, there is probably lots of waiting time [14:37:46] that's kind of important [14:38:07] how does webstatscollector differentiate? [14:38:09] https in the url? [14:38:18] no; it filters out internal ip adddresses [14:38:22] oh [14:38:35] it double counts things then? [14:38:50] no it never double counted. that's what we thought [14:39:00] oh i see, it doesn't count ssl reqs [14:39:15] because we we stopped sending the original stream [14:39:19] aye [14:39:20] and it's filter internal ip's [14:39:27] it doesn't' count it at all [14:40:02] so if we merge https://gerrit.wikimedia.org/r/#/c/63220/ then can ignore packet loss for ssl [14:40:13] and still have accurate monitoring for the rest [14:40:22] and accurate pageviews [14:40:25] WIN-WIN [14:42:02] hmm how many different udp2log loss metrics will this create then? [14:57:18] about 20 in ottal [14:57:57] on each udp2log host? [14:58:11] (i just commented on change) [15:13:08] drdee, re the kafka deb thread, I don't know what to say til I try it and start looking into it. [15:13:39] i think i don't understand this bit: [15:13:44] We should build the binaries from their respective sources, not embed and execute pre-made binaries. [15:13:48] (i mean, i understand it...) [15:13:49] but [15:14:08] ahh i'll just reply in thread… :)( [16:34:22] drdee: how did the hive job turn out? [16:34:55] milimetric: do you know the story with this "abandoned" change set https://gerrit.wikimedia.org/r/#/c/62792/ [16:35:06] looking [16:35:25] nope, never saw it [16:36:48] k [18:19:14] ok, drdee! [18:19:15] http://localhost:8888/filebrowser/view//wmf/raw/webrequest/webrequest-mobile/dt=2013-05-13_18.15.00/part-1366922903370_11638-m-00000 [18:19:22] the producer is running on an09 only [18:54:36] !log analytics1009 now sending geocoded anonymized mobile logs to kafka topic webrequest-mobile, this is consumed to /wmf/raw/webrequest/webrequest-mobile every 15 minutes [18:54:38] Logged the message, Master [19:04:33] New patchset: Diederik; "Added per dc / server role breakdown of udp2log packetloss monitoring." [operations/puppet] (production) - https://gerrit.wikimedia.org/r/63220 [19:04:40] ottomata, can you have another look? [19:04:56] this one keeps an overall packet loss metric without https, and per role per dc metrics [19:08:24] is udp-filter working well? [19:08:44] erosen, around? [19:14:32] erosen_ around? [19:14:35] ya [19:14:36] yup? [19:14:37] sup? [19:14:50] x to the c to the s? [19:15:06] word [19:15:14] trying to make progress [19:15:41] realized that it's not clear how the kraken counts should fit into things [19:15:46] looking, aaaaaaaaa, looks better, hmm [19:15:46] i still kinda think we should keep metrics named 'packet_loss_average' from each host [19:15:46] rather than all_roles: [19:15:46] ... [19:15:47] ganglia gets weird when you change metric names [19:15:47] can you do [19:15:50] the comparison has proven to not be that useful [19:15:55] adding to review... [19:15:55] drdee, comments inline [19:16:22] ottomata; didn't know about ganglia metric names stuff [19:16:24] yeah tell me how to do it! [19:21:57] ottomata: fixed! [19:24:29] erosen: soooooo to the batcave? [19:24:36] yeah [19:24:36] let [19:24:37] s [19:24:40] dd [19:24:41] o [19:24:42] i [19:24:42] t [19:27:53] cool drdee, i like that patchset, that way none of the existing stuff will break [19:44:28] drdee, if I put rolematcher.py in the same directory as PacketLossLogtailer.py, will import rolematcher work properly? [19:47:33] it should :D [20:47:12] so hey duuuuuudes, i've been out of it, what's the status with X-CS? [20:47:15] erosen, drdee? [20:47:23] yeeeaaas [20:47:31] we're hanging out [20:47:34] wnat to join? [20:47:35] oh! [20:47:36] ok [21:26:08] laters all! [21:26:52] drdee: http://reportcard.wmflabs.org/ [21:28:18] thx [21:55:00] guys, any cards that you want to see groomed tomorrow ? [22:20:14] drdee: check out the results in this card: https://mingle.corp.wikimedia.org/projects/analytics/cards/645