[13:44:51] hmmm *** stack smashing detected ***: ./udp-filter-static terminated [13:44:59] I get this on a "break" statement [13:45:07] on the very last line of the file [13:45:14] I'm still trying to figure out why that is [13:49:08] back to the incremental decommenting debug method I guess [13:52:50] this is an interesting method, commenting a big chunk and then incrementally decommenting parts of it until you find which part produces the problem [13:52:56] it's quite generic but it works every single time [14:01:45] ok found the problem [14:06:49] yo average_drifter, milimetric [14:06:49] drdee: good day ! :) [14:07:04] did you fix yesterday's segfault? [14:07:26] drdee: I found the cause and I know a solution too, but before I apply it I want to discuss a log line [14:07:31] first the link to the log line [14:07:37] and can you please update your asana tasks, close the ones that have been finished and indicate on which tasks you are working? [14:07:38] sure [14:07:41] what's the log line [14:08:02] https://gist.github.com/cc6d914483e7cc1c95d6 [14:08:14] the problem with it is the following [14:08:56] there are two ips in the x_forwarded_for field [14:09:16] drdee: I've sent them in pm to you [14:09:20] yup [14:09:30] the problem is that they're separated with ,%20 instead of just , [14:09:40] be two or more ups in log line is rather common [14:09:52] can't you split by comma [14:09:57] and then strip the %20? [14:10:21] those are actually 3 ips, but the first two of them are separated by %20 [14:10:27] instead of , [14:10:30] so it was overflowing a char array because of that [14:10:51] yeah I can split by comma, but the first two of them have no comma between them, instead they have %20 [14:13:17] I can fix that by stopping at the first % [14:13:23] if there is one [14:20:38] can you split by %20 as well? [14:23:24] yes, I just wrote code to stop at one of these conditions: [14:23:26] 1) comma is found [14:23:29] 2) % is found [14:24:08] 3) more than 39 characters (because the biggest ipv6 address is 39 chars long http://stackoverflow.com/questions/166132/maximum-length-of-the-textual-representation-of-an-ipv6-address ) [14:24:24] ok [14:33:04] morning [14:35:08] mooooooorning [14:36:40] :) [14:39:08] what's on your plate today? [14:41:03] hi milimetric , ottomata :) [14:41:37] hi drifter [14:41:56] :)) [14:42:49] hi! [14:42:55] morning ottomata [14:44:56] mooorning [14:47:26] ottomata, about the x-carrier header, we should actually wait until after the fundraiser, or not? [14:47:58] i think its ok, if they've deployed the x-carrier header, that means that we can starting filtering on it sooner rather than later, no? [14:48:04] filtering on x-carrier is only changes to our stuff [14:48:08] not stuff that other people use [14:48:43] well it would change the log format, it would be an extra fiedl [14:48:55] and that could affect the fundraising team [14:49:13] they didn't want us to add the language header either or replace the spaces with tabs [14:49:25] so i i guess it's the same issue [14:49:41] yes, that's true, but I guess I thought amit was saying it was already done [14:49:59] Patrick has deployed the X-carrier header. [14:50:27] but only on the mobile frontend [14:50:35] not in the traffic logs [14:50:52] so the x-carrier header is now available to us to write to a file [14:51:14] oh only for mobile [14:51:15] right ok [14:51:43] first week of january we should do all those changes in one big swoop [14:53:40] i worked yesterday with louisdang on oozie [14:53:57] we got a python script that generates an oozie workflow xml file [14:54:11] now figuring out how to use that in hue [14:54:35] cool! [14:57:02] so, all of the zero traffic is mobile (duh?) [14:58:12] yes [15:01:19] ok [15:02:10] ok, drdee, so this morning i'm going to finish some puppet stuff I was working on yesterday [15:02:17] aight [15:02:23] hopefully preilly will help me with x-carrier this afternoon [15:02:37] and, i want to talk to erosen about helping erik M with kraken stuff, i will set him up an account and such [15:02:45] but i think evan would be a better demoer at cool stuff you can do [15:02:56] and I will talk to evan about maybe jointly presenting kraken at metrics meeting [15:03:10] i can say: here is what it is! and evan can say: here is what I'm using it for! [15:03:22] eh? [15:05:35] sounds good, i spoke yesterday with erik [15:05:55] and i think it would be a real good idea if you guys would demo next thursday [15:11:17] ok cool, will do then! [15:31:56] ottomata: are $HOME/.ssh directories on stat1 managed by an external process ? [15:32:39] yes [15:32:44] by puppet [15:32:49] currently when I build a binary I do build on build1 => scp binary locally => scp to stat1 [15:33:01] and I wanted to copy the binary directly build1 => stat1 [15:33:24] yeah, that's a networkign thing, its really annoying [15:33:25] i don't' think that's possible [15:33:34] they keep labs and prod machines separate [15:33:40] hmm ok [16:46:37] * average_drifter is away for 40m to get stuff before store closes [17:13:50] i think that sounds great, otto [17:47:45] drdee, we were asked not to make changes to the logging sources [17:48:02] not how we collect logs using udp-filter [17:48:25] but the x-carrier change is a change to the source, right? [17:48:30] it is already done [17:48:32] by patrick [17:48:38] amit's message says he deployed it [17:48:48] yes in varnish on the frontend [17:48:50] Andrew- [17:48:50] Patrick has deployed the X-carrier header. [17:48:57] that's the logging source, no? [17:49:02] so its already been changed [17:49:04] no [17:49:21] no? [17:49:27] that's on the frontend detecting mobile zero traffic and displaying a banner [17:49:38] the x-carrier does not yet show up in the udp stream [17:49:55] and fundraiser asked us not to change the udp log until after the fundraiser [17:50:02] hm [17:50:02] so [17:50:07] x-carrier is set by the mobile provider carriers [17:50:12] no [17:50:16] we set it [17:50:41] what did patrick do with x-carrier then? [17:51:19] won't be at scrum today, fyi--GLEE (Global Grantmaking, Learning and Evaluation Enterprise) meeting [17:51:26] k [17:51:27] still working on mobile dashboard [17:52:23] so, varnish sets the x-carrier http header based on the ip range check and displays a banner telling the viewer that she is using wikipedia zero [17:52:40] what's actually shown depends on the contract with the carrier [17:52:54] so we *could* start logging the x-carrier header now, it's available [17:53:01] ah, erosen, I want to talk to you about metrics demo [17:53:02] but that means we change the log format, right [17:53:11] ya [17:53:16] got time after your meeting? [17:53:16] and that is likely to break fundraising scripts doing analysis [17:53:18] i've been meaning to respond to that [17:53:24] sort of [17:53:30] checking [17:53:42] soooo, patrick just set the x-carrier header, which was not happening before? [17:53:43] that's why they asked us not to make any changes to the log format right now [17:53:50] ottomata: does 1130 work? [17:53:51] hangout? [17:54:08] patrick made it so that x-carrier is now a 2 letter acronymm [17:54:09] i have my 1:1 with robla then [17:54:17] it used to be the whole name of the carrier [17:54:36] erosen, after my 1:1, i guess at 1200 your time? or maybe after you lunch? [17:54:41] hm, ok [17:54:42] this was and old concern of binasher that the full carrier name would generate too much traffic [17:54:46] so yes we can fix it now [17:54:51] 12pst is fine [17:54:55] and I really want to fix it [17:55:04] but the fundraiser........ [17:55:06] aye [17:55:08] ottomata: will schedule [17:55:18] cool [17:58:13] https://plus.google.com/hangouts/_/2e8127ccf7baae1df74153f25553c443bd351e90 [19:01:40] ottomata, trying to push wikistats from stat1 but it's asking me for my credentials [19:10:56] ottomata ^^ [19:19:56] drdee: hm, what's the path on stat1? [19:20:14] stat1:/a/wikistats_git/ [19:20:21] i used to be able to push [19:20:45] but i think you changed the url [19:20:49] to gerrit [19:20:52] not sure though [19:21:49] hmmmm, i think I switched it to the readonly url [19:21:51] so you could both pull [19:21:59] the write urls have usernames in them [19:22:06] really [19:22:08] if you want to push [19:22:13] you should have a copy in your home dir that you clone [19:22:15] and then you push that [19:22:19] and then if you have a globally deployed place [19:22:25] you can pull the readonly (non user specific) one from there [19:30:16] ottomata: running a little late. I'd like to make sure we're on track to fix an itwikisource issue before starting [19:31:24] hokay! [19:31:31] i have a little meeting with erosen in 30 [19:31:36] he and I could do that now instead [19:32:07] ya [19:33:17] https://plus.google.com/hangouts/_/57b9f21b30beca32d45e87bee316ce14b388decc [19:33:19] drdee, join us? [19:33:50] yes [19:33:54] one me way [19:33:57] one sec [19:33:58] k [19:58:58] ottomata: I'm finally done now [19:59:23] if you're still meeting, I can go grab a quick bite and be back in 10min-ish [19:59:24] ok cool [19:59:27] naw we're done, can do [19:59:32] but 10 mins is fine to [19:59:33] either way [19:59:38] let's meet now [20:00:19] mk [20:57:30] erosen [20:57:30] http://www.mediawiki.org/wiki/Analytics/Kraken/Access [21:10:48] yo [21:10:51] sup? [21:10:56] oh yeah [21:10:57] the wiki [21:10:59] one sec [21:11:01] in a meeting [21:12:53] no probs, just add some links to some good pig tutorials maybe? [21:14:53] ya [22:47:53] ottomata1: updated http://www.mediawiki.org/wiki/Analytics/Kraken/Access [22:48:04] danke! [22:48:09] maybe we don't need the example script, so feel free to remove it [22:50:31] no that's awesome [22:50:32] thank you [22:50:36] i just gave erik m the link [22:51:57] cool [23:43:07] erosen, around?