[00:03] lifeless, hey, still around? [00:04] yeah [00:05] lifeless, I have a minute or two. What's up? [00:08] lifeless, we had a bunch of discussion today about the qa-shepherd thing - cut the scope down again, it should be ready sooner than what we expected before [00:10] lifeless, now that we have done that, I can send you that email update you asked for [00:11] mars: I was offering to dive in and help [00:12] mars: rather than a status update [00:12] lifeless, well, now we have the scope down to where the three of us have to worry about stepping on each-other's toes :) [00:13] nice [00:14] lifeless, thank you for the offer, I'll keep it in mind. I think we'll be fine for the first few days of next week. [00:14] ok cool [00:15] (I don't do 'status update requests' unless I -really- can't help it - just for clarity :) [00:15] lifeless: shouldn't you be on saturday? [00:15] yes [00:15] lifeless, makes sense :D [00:15] mtaylor, it's not Saturday here, and he asked first :) [00:15] weird PPA build issue- only shows up on i386 lucid builder: [00:15] dpkg-deb: parse error, in file 'debian/drizzle/DEBIAN/control' near line 6 package 'drizzle': [00:15] `Depends' field, reference to `drizzle-server': version contains ` ' [00:16] mars: hehe. his fault then [00:16] mars: so generally the fastest way to make something go faster is to help :- which is why I offered ;> [00:16] http://launchpadlibrarian.net/53201197/buildlog_ubuntu-lucid-i386.drizzle_2010.08.1683-1ubuntu1~lucid0_FAILEDTOBUILD.txt.gz in case anyone wants to look [00:23] mars: still around? [00:23] lifeless, yep, thinking [00:24] mars: so if there isn't anything sensible I can hop in and do; don't bother mailing me ;) [00:24] lifeless, sound fair to me. [00:25] sounds even [00:25] need to run again [00:26] lifeless, I'll ping you back with specifics if I get a moment back at the keys [00:44] lifeless, at the keys, if you are still there, skype? [00:48] mars: hi [00:48] lifeless, yep? [00:49] I amhere [00:49] skype? [00:49] it's faster, trust me [00:53] lifeless, https://code.edge.launchpad.net/~ursinha/qa-tagger/setuptools-conversion [00:54] lifeless, https://code.edge.launchpad.net/~launchpad-qa/qa-tagger/devel [01:13] lifeless, http://bazaar.launchpad.net/~launchpad-qa/qa-shepherd/devel/annotate/head:/shepherd/testing.py [01:17] https://bazaar.launchpad.net/~ursinha/lp-qa-tools/lplib-fakes/files [01:19] make check working for me now :) [01:24] lifeless, btw, feel free to move the cards on the board if you get a chance to hack on something. That way we can pick or pass the work on Monday morning. [01:25] "we" being those of us on the opposite side of the planet from you :) [01:38] :) [23:26] How can you tell when its more efficient to do separate queries and when its more efficient to just do a single query with joins [23:26] ? [23:27] the science of measurement [23:27] also some context will help [23:27] there are places where its always better to do it one way or another. [23:28] e.g. per-object queries in an iterator for an api or web page -> always want one query ('build the table of data to show') [23:28] this is a scaling question - ask yourself what happens when you have BigN (e.g. 400 people on one page) [23:29] I was just wondering if there was a quick and easy way to tell... like look at the output of EXPLAIN and look for this or that. [23:29] When's it going to be better to do multiple queris? [23:32] re: buildd-master failure, 2010-08-07 23:26:24+0100 [-] exceptions.AttributeError: 'NoneType' object has no attribute 'email' [23:33] cody-somerville: where do you get access to that ? [23:33] escalating to phone. [23:33] cody-somerville: Thanks. [23:34] cody-somerville: Is there something before that indicating which build it's trying to send a failure message about? [23:34] wgrant: different domain lookups that aren't related to each other [23:34] 2010-08-07 23:26:24+0100 [-] Starting templates build trunk-3771906 for lp:docky. [23:34] 2010-08-07 23:26:24+0100 [-] startBuild(Lucid, daily, gnome-terminator (on http://doubah.ppa:8221/)) [23:34] 2010-08-07 23:26:24+0100 [-] Scanning failed with: 'NoneType' object has no attribute 'email' [23:34] lifeless, devpad [23:35] Aha. [23:35] cody-somerville: where - /srv/ somewher e? [23:35] * wgrant prepares the SQL. [23:35] lifeless, /srv/launchpad.net-logs/soyuz/cesium/buildd-manager.log in this case, yea [23:35] cody-somerville: thanks [23:36] elmo will be around shortly [23:36] [23:36] wgrant, would having a launchpad admin set a preferred e-mail address for the requester work? [23:36] cody-somerville: It might, but we could also just suspend the build to preserve the evidence. [23:36] suspend it please [23:36] we need to fix root causes [23:39] I've ran into this issue myself with scripts that use the launchpadlib [23:39] I had assumed that every user had preferred email address I could grab. [23:39] elmo: SELECT * FROM buildqueue, job, branchjob WHERE buildqueue.id = 3771906 AND job.id = buildqueue.job AND branchjob.job = job.id; [23:41] Part of the issue is that you have to get the actual e-mail address from an attribute on the user that doesn't exist if there is no preferred e-mail address available [23:41] (I really don't know why a translations job dispatch would be trying to acesss an email address) [23:41] wgrant, its a recipe build thats causing it to die [23:41] wgrant, line 66 lib/lp/code/model/recipebuilder.py [23:42] Oh, recipe build logging is fucked. [23:43] lifeless, are you going to own the incident report? [23:44] cody-somerville: thanks for volunteering :) [23:44] elmo: You should have a cancel button on https://code.edge.launchpad.net/~gnome-terminator/+recipe/daily/+build/515 [23:44] lol, alright. :P [23:45] man, seriously [23:45] this is Ng's fault? [23:45] its a bug in translations somewhere [23:45] No, it's not. [23:45] The logs are just misleading. It's actually a recipe build. [23:45] wgrant: I do? [23:45] wgrant: sorry [23:45] wgrant: I do. hit it? [23:45] wgrant: oh, grah [23:45] elmo: Please. [23:46] wgrant, the recipe build belongs to Ng [23:46] anyhow, bug, not user error, right ? [23:46] lifeless: The first two lines of that log that cody-somerville pasted are equivalent, but for different job types. Yay for consistency. [23:46] cody-somerville: It is, yes. [23:46] lifeless: Right. [23:46] wgrant: done [23:46] wgrant, sorry, misread you [23:47] We're 20 seconds in... let's see if it sticks. [23:48] Apparently not. [23:50] Is there another build appearing in the logs now? [23:50] Ah. [23:51] The other two builds are now at the head of the queue. [23:51] 2010-08-07 23:50:38+0100 [-] startBuild(Maverick, daily, gnome-terminator (on http://muntries.ppa:8221/)) [23:51] 2010-08-07 23:50:38+0100 [-] Scanning failed with: 'NoneType' object has no attribute 'email' [23:51] https://code.edge.launchpad.net/~gnome-terminator/+recipe/daily/+build/516 and https://code.edge.launchpad.net/~gnome-terminator/+recipe/daily/+build/517 need destruction. [23:51] Yep. [23:51] done [23:53] 40 seconds in. Looking good. [23:53] I'm tailing the log looking for noe [23:53] noNe [23:53] will this happen again tomorrow? [23:54] Possibly. [23:54] awesome [23:54] The API won't tell me who requested them :( [23:55] Hmm. [23:56] I guess it probably uses the recipe owner for daily builds. [23:56] So it might be a good idea to disable daily builds of that recipe for now, or it will indeed happen again tomorrow. [23:56] do you know the url for that offhand? [23:57] or even pointers to it [23:57] https://code.edge.launchpad.net/~gnome-terminator/+recipe/daily/+edit [23:57] There should be a checkbox. [23:57] WTF [23:57] that says *I'm* the owner [23:57] Oh no, not that bug... [23:57] that's either an awesome troll by Ng or an epic UI bug in LP [23:57] It's a known bug, yes. [23:58] Either find someone in the team to do it, or just twiddle the flag manually. [23:58] manually == SQL? [23:58] Yeah. [23:58] * elmo rolls his eyes [23:58] Sorry :( [23:58] it's not your code [23:58] (I assume) [23:59] Heh, no. [23:59] And we have logtails... so everything's working again for now. Thanks.