[16:01:47] <bearloga_>	 good morning halfak lzia guillom Ironholds! :D
[16:01:54] <guillom>	 hello hello
[16:01:54] <Ironholds>	 morning bearloga_ !
[16:02:00] <Ironholds>	 can you answer my dwell-time question now? :P
[16:02:09] <lzia>	 good morning bearloga_ :D
[16:02:12] <lzia>	 wasup?
[16:02:22] <lzia>	 Ironholds: hiii
[16:02:25] <lzia>	 question for you Ironholds
[16:02:26] <lzia>	 :D
[16:02:29] <Ironholds>	 lzia, sure!
[16:02:51] <bearloga>	 Ironholds: sure!
[16:02:54] <lzia>	 your favorite topic: did you compare sampled log pageview counts with actual counts based on the logs?
[16:02:54] <Ironholds>	 bearloga, danke!
[16:03:11] <Ironholds>	 lzia, when?
[16:03:23] <lzia>	 I'm trying to understand how far off sampled log trends can be from the actual log counts
[16:03:33] <lzia>	 at any point Ironholds
[16:03:44] <Ironholds>	 lzia, oh, yes. I've got a writeup, hangon
[16:03:51] <lzia>	 oooo, nice, Ironholds!
[16:04:09] <Ironholds>	 lzia, http://ironholds.org/blog/content/images/2015/08/sample_testing.png
[16:04:23] <Ironholds>	 I would do a full report on this but that's Not Mah Job :D
[16:04:33] <Ironholds>	 but basically you're looking at teeny tiny amounts of variation
[16:04:40] <Ironholds>	 like, most of the time it's hard to tell those are different lines
[16:05:13] <Ironholds>	 it gets more problematic if you compare to the historical sampled data in the Pentaho cubes because then you run into temporal issues - one implementation getting patched before or after the other to deal with $weird_edgecase
[16:06:10] <lzia>	 Ironholds: this is helpful
[16:06:18] <Ironholds>	 awesome
[16:06:56] <lzia>	 one question: sampled log on this plot is sampled log count times 1000?
[16:07:58] <Ironholds>	 I hope so otherwise our data is truly screwed!
[16:08:02] <Ironholds>	 (yeah, it is)
[16:08:27] <lzia>	 okay, great.
[16:08:31] <lzia>	 thanks, Ironholds.
[16:10:20] <Ironholds>	 no problem!
[16:36:47] <Jeph>	 Is the watchlist table available on the toollabs db slaves?
[16:37:24] <Jeph>	 or is it called something else?
[16:37:26] <Ironholds>	 I do hope not
[16:37:38] <Ironholds>	 if it is it'd be, I guess, t_watchlist or watchlist_t but I really do hope that data isn't there.
[16:39:27] <Jeph>	 I can't find any of them, so they arn't publicly available?
[16:40:31] <Ironholds>	 Jeph, apparently not
[16:44:37] <Jeph>	 On a diff note, select page_title,page_id, YEAR(rev_timestamp),MONTH(rev_timestamp), count(*) from page join revision_userindex on page.page_id = revision_userindex.rev_page where page_namespace=0 and page_is_redirect=0 and rev_user !=0 and lower(CONVERT(rev_user_text USING latin1)) not like '%bot%' group by YEAR(rev_timestamp), MONTH(rev_timestamp), rev_page having count(*) > 4;
[16:44:53] <Jeph>	 is a query I've never been able to run for 'en'
[16:45:06] <Jeph>	 it always timesout
[16:46:02] <Jeph>	 I set both : interactive_timeout=288000  & wait_timeout=288000
[16:46:35] <Jeph>	 any help will be much appreciated :-)
[16:49:51] <halfak>	 Yeah.  No indexes helping you there. 
[16:50:06] <halfak>	 It's better to find bots with the user_groups table
[16:59:13] <halfak>	 Regretfully, I don't have time to hack on this right now. 
[16:59:37] <halfak>	 query optimization is one of my fav. things to do in this channel, but I've got some paper deadlines coming up on top of my regular work :/
[17:02:34] <Jeph>	 so I should get rid of the string check and check the user table instead?
[17:03:09] <Jeph>	 I'll try that, any other quick observation for me?
[17:03:24] <Jeph>	 without taking too much of your time :-)
[17:45:11] <halfak>	 Jeph, sorry.  I'll try to have a look later. 
[17:53:33] <Jeph>	 Sure, thanks :-)
[20:07:28] <yuvipanda>	 wah, look at this madhuvishy 
[20:07:35] <yuvipanda>	 getting puppet patches merged left and right :P
[20:07:39] <madhuvishy>	 he he
[20:07:46] <madhuvishy>	 this was analytics