[14:12:14] morning guys [14:17:41] morniing [14:23:51] Morniiing [14:31:55] yo qchris [14:32:06] Hi drdee [14:34:33] ottomata: Found you! [14:34:42] ottomata: About accessing kafka-main1 ... [14:34:43] ah ha! [14:34:50] yeah man! try kraken-namenode-standby [14:34:53] ottomata: Can you access it? [14:34:55] Oh ok. [14:34:56] thats' where i do my camus stuff [14:35:09] yes I can access it [14:35:11] both of them [14:35:14] I cannot :-( [14:35:19] Permission denied (publickey). [14:35:30] I have that a lot of labs instances [14:35:35] Mostly it's temporary [14:35:40] But here it is not :-( [14:36:04] Can there be problems with my home directory? [14:36:21] hm [14:36:33] looks ok to me [14:36:51] ssh -v kraken-namenode-standby.pmtpa.wmflabs [14:36:51] So labs just does not like me :-) [14:36:58] paste output somewhere [15:22:09] hi! [15:22:37] apologies for the radio silence this morning [15:34:29] morning [15:50:27] milimetric: i'm running your count query on 2 days worth of varnishkafka data from those 2 hosts [15:50:42] 450G [15:54:13] cool ottomata, hope it's clean [15:59:48] milimetric: hey [15:59:56] drdee: hi [15:59:57] howdy [16:00:41] was wondering if Python has some sort of import guards [16:01:06] similar to the classic #ifndef __MODULE #define __MODULE class_definition #endif [16:01:14] that you see usually in C or C++ for structs or classes [16:01:29] I'm bumping into a circular import again [16:01:36] I'm in the batcave if you wanna talk to me [16:01:44] I can show you what I mean [16:01:46] no python does not have that [16:02:08] you should put the circular dependencies in a new parent module to break the cycle [16:02:09] ok [16:03:06] average: you can of course do if something: import something, but that statement would only run once when the file is originally imported [16:03:16] so your import would not work [16:03:27] there are three solutions: [16:03:37] 1. obviously, rework your imports so there's no cycle [16:04:07] 2. move the import that's causing the cycle closer to where that import is actually needed (ie inline with the execution) [16:04:34] 3. do something fancy like we did with configurables and the wikimetrics __init__ [16:05:19] in that __init__, we make sure the order of the imports is guarded by the __init__ and should fix circular imports for the whole project [16:05:33] coming to the batcave though, one sec [16:18:47] milimetric: drdee [16:18:48] hostname missing_sequence_runs [16:18:48] cp1048.eqiad.wmnet 0 [16:18:48] cp3003.esams.wikimedia.org 0 [16:18:48] Time taken: 1780.87 seconds [16:18:49] :) :) :) [16:20:25] nice [16:20:28] kinda took a while [16:21:50] yeah 30 mins [17:03:07] ottomata: scrum [17:03:45] boop [19:04:09] ottomata: quick chat about how to go forward with importing the hourly files? [19:04:54] sure! [19:05:39] batcave [19:05:54] omw [19:22:08] milimetric: https://github.com/spotify/luigi [19:23:24] yes yes [19:23:26] i like luigi [19:26:58] hey -- what's the consensus on the internal mailing list name? [19:27:59] tnegrin: I always use: analytics-internal@lists.wikimedia.org [19:28:11] That's what mailman declares as authorative: [19:28:16] https://lists.wikimedia.org/mailman/listinfo/analytics-internal [19:28:47] I see both in my address book :( [19:28:56] I'd like to delete the other list entirely [19:29:19] Mailman does not know about the other one. [19:30:34] mil imetric provided a link to clean your autocomplete from the addressbook: [19:30:38] http://www.timeatlas.com/email/gmail/cleaning_your_gmail_autocomplete_list#.UlLmNJyVJ4x [19:30:52] Does that work for you? [19:36:47] sure -- I'll just do that [19:37:54] Splendid! Thanks milimetric for finding that link. [19:38:15] totally -- that was a really weird problem [19:38:27] just goes to show I shouldn't be allowed to operate computers [19:38:29] ;) [20:12:23] milimetric: ?