=== _mup__ is now known as _mup_ [08:59] hi gmb: how are you doing? [09:00] frankban: I'm good, thanks. Did I miss anything yesterday that I need to be aware of? [09:00] gmb: not if you've red the handoff emails by gary [09:01] gmb: could you take a look at https://code.launchpad.net/~frankban/launchpad/bug-980295-soyuz-upload/+merge/102255 [09:01] > [09:01] ? [09:01] Not yet; I'll look at them presently. [09:01] frankban: Sure; I'll look at that in a few minutes. [09:02] gmb: after that, what do you think about pairing on something? We are in the subunit/testr/buildbot land and you know how things are tricky there [09:03] frankban: Ok, sure. Let me get up to date on everything and I'll ping you when ready [09:04] gmb: thanks, great [10:06] frankban: I've approved your branch. What is it that we need to pick up next, then? [10:08] gmb: thanks. we can decide looking at gary's email. brad is working on subunit. maybe we can pick buildbot, or something else from the kanban board [10:09] Ok. [10:09] Looking... [10:15] frankban: "buildbot results show subunit streams filtered by workers" looks interesting, though I'm not sure we can do all that much with it until the bug 949950 work lands. [10:15] <_mup_> Bug #949950: Feature request: an option to show full subunit stream of running tests < https://launchpad.net/bugs/949950 > [10:17] gmb: looking [10:20] gmb: it's described by gary as: [10:20] * buildbot's testr call in master.cfg should get the --subunit argument from the work on bug 949950 [10:20] * we add multiple stdout parsers in the buildbot output for the testr call. [10:20] > One will simply be the stdout. [10:20] > We will have one for each of the test processes piped through subunit-filter to give us the output per worker (e.g. subunit-filter --with="^tags: worker-0" -error --failure --success), and then piped through "subunit-ls --no-passthrough" [10:20] <_mup_> Bug #949950: Feature request: an option to show full subunit stream of running tests < https://launchpad.net/bugs/949950 > [10:20] > We will have one that will show us only the failures, perhaps in non-subunit format. This one might also update the status of the waterfall for on the fly test counts. [10:20] * for the buildbot piping to work, we *might* need to clean up the output of our testr command (i.e., removing lines 1-44 of http://paste.ubuntu.com/932523/). it might be robust enough without it. [10:20] gmb: I don't know if we need the subunit output from testr to use subunit-filter [10:21] Hmm. [10:21] gmb: to test it we could take the output from the .testrepository/tmpxxxx file, maybe [10:21] frankban: Yes, that's what I was thinking as well. [10:23] gmb: I have started the charms, using constraints (hope they work) [10:24] frankban: Okay. I'm still waiting on being able to kill VMWare - it's eating my CPU but I'm in the middle of apt-get upgrade on the VM; I'll try and start a hangout but no promises. [10:36] aarghh... restarting the juju environment --constraints mem=10G --constraints cpu=8 gave me 4 cores... [10:36] ! [10:37] I love Juju's ability to be completely and utterly maddening at almost every turn. [10:37] frankban: I'm in https://plus.google.com/hangouts/_/extras/canonical.com/hallo-tuesday#; no idea how good my audio will be though. [10:39] Aaaaah [10:39] Firefox crash. [10:39] Thankfully, I'm nearly done with the updates... [10:40] Restarting FF, anyway... [10:42] ok, constraints don't work... [10:46] frankban: Oh, fun. So am I right in thinking that there's now no way to start the kind of instances we need? [10:46] gmb: I am trying to check if the error is in the juju wrapper, using the real juju now [10:47] Ah, okay. [10:47] gmb: if that's not the case, I could use constraints in environment.yaml [10:48] frankban: Ah, right. For some reason I thought that wouldn't work - I guess I was confusing it with default-instance-type and suchlike, which are now non-functional. [10:48] or maybe... gmb do you know a way to specify the ec2 image type in contsraints? [10:49] Hang on, I think bac wrote some useable instructions... [10:49] frankban: http://pastebin.ubuntu.com/927016/ [10:50] That worked for me last week, IIRC. [10:50] So I guess tweaking that for a c1.xlarge instance should work. [10:50] "should" [10:51] cool thanks gmb [10:51] frankban: If you're starting an lpbuildbot master/slave, it probably makes sense to set that running and then grab lunch. By which time, I should also have my CPU back. [10:52] gmb: ok [10:52] see you later [10:52] Okay. [10:53] frankban: I've added our faces to the card for this on the Kanban board. [10:53] thanks gmb [10:58] gmb: your ssh key is the carrot one? [10:58] frankban: Yes. [11:03] gmb: added your keys and started 2 screen sessions [11:03] master: ec2-204-236-222-6.compute-1.amazonaws.com [11:03] frankban: Thanks. [11:03] slave: ec2-184-73-121-132.compute-1.amazonaws.com [11:03] Okay. I'll grab food and check them out later. [11:12] hello gents [11:12] carrot keys? [11:13] gmb and frankban: I'm readying email and looking at the board to figure out what I should work on. Is the board up to date? Do you have any suggestions? [11:13] hi benji -- i think it is euro lunch [11:14] bac: "euro lunch" sounds like a new fad, one I could get behind [11:15] benji: yes. euro* maybe not but *lunch definitley [11:15] :) [11:16] benji: did you see gary's emails re: the testr/subunit/testtools stuff? [11:16] yep, I'm re-reading them now to try to figure out where I can be most helpful [11:16] he and i are in the middle of trying to get the --subunit option to work [11:17] yep [11:18] bac: unless you want to pair, it looks like I should work on one of the prerequisites for 974622 [11:20] benji: maybe we can pair later? pairing before 9am is probably a bad idea [11:20] heh [12:07] I was going to restart after that upgrade, but running out of time [12:08] benji, how did the manuel py3 work go? [12:08] ooh, nvidia app got a nice new pretty icon [12:08] bac benji frankban gmb, call in 2 [12:08] gary_poster: really well, we had the tests passing around noon and had a release made by 3 (with some Thai in between) [12:09] benji, great! :-) [12:09] I suspected Thai might play a part in the day [12:10] umm, thai [12:10] If my magic trackpad worked reliably, I would not have a single complaint about the more recent precise versions. [12:31] bac, https://talkgadget.google.com/hangouts/_/extras/canonical.com/goldenhordepart2thequickening ? [12:31] gary_poster: ok [12:47] gary_poster: what should I look at now? [12:48] benji, good question. bac and I are talking in https://talkgadget.google.com/hangouts/_/extras/canonical.com/goldenhordepart2thequickening . I need to prepare and deal with kids very very soon... [12:49] benji, could join us there for now [12:49] gary_poster: is that an invitation [12:49] oh, it is [13:00] rejoining [13:44] bac: good to know: "gnome-open" is roughly equivalent to OSX "open" [13:52] benji: here is what happens when i exit "script" while sharing a screen: http://paste.ubuntu.com/933977/ [13:53] something got terminated [1] but not the ssh to chinstrap as you see it is still running [13:54] bac: hmm, I don't understand that [13:54] me neither [14:01] bac: we want these in lp:~yellow, right? [14:02] benji: yes, i think so [14:10] gary_poster, benji: here is the final diff: http://pastebin.ubuntu.com/934005/ [14:10] bac: thanks; the testtools roll-up branch is ready at lp:~yellow/testtools/latest [14:11] ok [14:11] still working on MP [14:11] k [14:11] nice bac [14:19] benji: my branch is at lp:~bac/testrepository/bug-949950 [14:20] k [14:20] benji, bac, it looks like we have basic approval from Robert on https://code.launchpad.net/~yellow/testrepository/integrate_worker_tagging/+merge/102165 [14:20] jml will adjust the wrap_results approach per his review [14:20] great [14:21] and then we can write a test and try to get it officially merged [14:21] ISTR not being quite sure what we could test; looking... [14:21] heh [14:22] yeah, if we could test to see that "case" had the expected wrap_result, and that wrap_result did what we expected, that would be an acceptable unit test, but case is not exposed directly... [14:23] we could write wrap_result in such a way that it could be tested in isolation... [14:24] that might be sufficient, though IWBNI we could verify that it was being used in the proper integration [14:25] benji: do you want to hangout to do the ppa recipes? [14:26] bac: sure, if you want to get it going I'll be there in a minute when I get this other roll-up branch done [14:26] benji: ok, the google won't let me into the hangout. :( [14:28] frankban, could you let me know what version of lxc is installed in ec2? 0.7.5-3ubuntu51 is supposed to make the manual fstab change unnecessary, so I'm wondering if that has propagated yet [14:28] all the way to ec2 [14:28] (which is sometimes behind the times) [14:28] no rush, but today would be good [14:29] ok, lp:~yellow/testrepository/latest is ready now, too [14:30] gary_poster: 0.7.5-3ubuntu52 [14:30] frankban, uh-oh [14:30] bac: try it now (https://talkgadget.google.com/hangouts/_/extras/talk.google.com/goldenhorde#) [14:30] frankban, we should verify that the manual hack is still necessary then [14:30] hmm, that's not the canonical one, but I guess it should work [14:30] if it is, we should report it [14:30] and maybe install our workaround in setuplxc [14:31] frankban, once you have the current test run, do you have the time to undo the manual change you made and rerun the tests? [14:32] gary_poster: one test run actually automatically started and finished without the hack: http://ec2-204-236-222-6.compute-1.amazonaws.com:8010/builders/lucid_lp/builds/0/steps/shell_9/logs/stdio [14:33] frankban, ack, thanks. Yeah, looks like the hack is still necessary. I'll consult with hallyn [14:34] gary_poster: meanwhile I am going to start a test with the hack [14:34] cool thanks frankban [14:34] benji: it was too much fun [14:34] benji: want to try the normal hangout? [14:35] bac: sure, do you want to make one or shall I? (I can't make a "canonical" one for some reason) [14:35] benji: i'm in [14:35] yeah, it is messing up today wrt to identities [14:36] bac: what is the URL? [14:36] you know, the normal one [14:36] https://talkgadget.google.com/hangouts/_/extras/talk.google.com/goldenhorde# [14:36] bac: "We're sorry, you are not invited to that hangout" [14:36] gah [14:42] frankban, I spoke with hallyn and apparently this is expected [14:42] we should adjust setuplxc [14:42] to make this change [14:43] eventually there will be an official fix for lucid containers [14:43] but not right now [14:43] i'll make a card [14:44] gary_poster: add an option to setuplxc to add that line in var/lib/lxc/lptests/fstab? [14:45] frankban, yeah. I guess an opton makes sense so we can easily undo it in the future, but for now it is required, so you could make the change without an option too [14:46] gary_poster: it could make sense to add the hack to lpsetup too [14:46] frankban, agreed [14:49] * gary_poster babysits/lunches quite early [15:19] parallel tests always find horrible ways to fail... [15:27] sometimes it's like working with quantum computers: http://dilbert.com/strips/comic/2012-04-17/ [16:08] frankban, :-) . Do you have new results yet? I didn't see them on the ResultsLog page [16:09] gary_poster: I am looking at them right now, some new errors: http://ec2-204-236-222-6.compute-1.amazonaws.com:8010/builders/lucid_lp/builds/1/steps/shell_9/logs/stdio [16:17] So a DisconnectionError, some weird hwdb thing, another DisconnectionError that looks like bug 974617 (we thought it was fixed), and our old friend UncleanReactorError [16:17] <_mup_> Bug #974617: test_operational_view_integration fails intermittently in parallel tests

< https://launchpad.net/bugs/974617 > [16:19] frankban, could you get the output of testr failing for the wiki's pastebin? I think that is easier to read [16:19] frankban, I'll file bugs, unless you want to [16:20] thanks gary_poster, I will update the ResultsLog page [16:20] cool thanks frankban. I'll put the bugs I file there [16:30] gary_poster: I hope the UncleanReactorError will be a dead old friend tomorrow: could you suggest me a place in launchpad where to put some testing generic tools, like the clean_up_reactor function? [16:31] frankban, ...lp.testing? looking there [16:32] frankban, yeah, lib/lp/testing/__init__.py? look in there and see what you think [16:33] gary_poster: yes I was looking there too, it seems a good place, thank you [16:33] cool, welcome [16:38] darn: "Rejected: testrepository_0.0.5+z~ppa1~precise1.dsc: Version older than that in the archive. 0.0.5+z~ppa1~precise1 <= 0.0.5++trunk146~yellow2~precise1" [16:39] heh [16:39] opaque to me [16:42] I think we need that to be "++z"... yep, that works [17:30] bac: testtools and testrepository are happy now (https://launchpad.net/~yellow/+archive/ppa) [17:33] well, now that all the testtools branches are merged we should switch to using the testtools trunk and fix our testr branch, I'll do both of those [17:55] benji: could you change the ownership of the testr build recipe to be ~yellow? [17:56] benji: also, in the past we had different build recipes for different branches (yellow branch, trunk, etc). i think that is wrong and confusing. we should have one recipe and just change the branch that is used. [18:03] hey gary_poster did you see jml's comments on my MP? https://code.launchpad.net/~bac/testrepository/bug-949950/+merge/102317 [18:03] bac, no, looking [18:04] bac: (was afk) yep, I'll change ownership now; on the second part (different build recipes for different branches), I don't understand what action you're wanting me to take [18:05] ownership changed [18:05] benji: encouragement not to create a new recipe for testtools trunk [18:06] bac: oh! right, I was going to edit the recipe we made to point it to the trunk, is that what you are wanting? [18:06] benji: yes, as opposed to what stupid people do (me) -- create a new recipe [18:08] bac, comments: sounds ok. (1) I thought we saw all tests when we ran the subunit make check with --subunit? I'm mistaken? (2) changing make_result as he describes: :-/ sounds like fixing the tests will be time-consuming and annoying. Hopefully I'm wrong. "vars" comment: I wonder if he's asking for a change to getattr. Not clear. :-/ [18:09] recipe updated, builds requested [18:09] gary_poster: as to 1) i'm not sure. [18:09] 2) yes, i think it'll break lots [18:09] :-/ [18:09] 3) using getattr looks more standard [18:10] true [18:10] gary_poster: would you grab the branch and see if the output is what you remember? [18:10] sure bac. gimme a sec, in the middle of something else, but can do in 5 [18:11] thanks [18:14] bac, no I remember lots of blather [18:14] maybe those were just the failing tests [18:14] but I thought it also included successes [18:16] bac, are you working on that? [18:16] gary_poster: i don't think we changed anything significant related to that [18:16] I don't think so either bac [18:16] gary_poster: yes [18:16] cool bac thank you [18:16] gary_poster: but i wanted to see if we had a regression before i implemented jml's suggestions [18:17] I doubt it [18:17] I suspect my memory/care in looking atthe result yesterday more [18:18] gah! since the recipes evaluated to having the same version the new builds failed, I think I can fix that without too much trouble [18:18] (I should have changed ppa1 to ppa2.) [18:19] benji: i sure wish the build machinery tested the versioning before doing all of the work... [18:19] really [19:57] gary_poster: i think we confused ourselves yesterday when we ran: [19:57] python -m subunit.run testrepository.tests.test_suite [19:57] that does show all of the results [19:57] think that could've been it? [19:58] bac, oh, yeah, maybe so [19:58] gary_poster: anyway, i've about finished the follow-up branch [19:58] test fallout was not so bad [19:59] bac, awesome. thank you [20:03] gary_poster: better, no? http://pastebin.ubuntu.com/934507/ [20:04] bac, yes :-) [20:06] I'm heading out to the dentist. I'll see you guys tomorrow. [20:06] here's the diff gary_poster: http://pastebin.ubuntu.com/934512/ [20:06] bye benji -- good luck [20:06] bye benji [20:09] bac, looks nice. you need to add the options for the load command too, according to jml, right? Did I miss that? [20:21] gary_poster: oh yeah [20:43] gary_poster: that was easy. MP created. [20:44] bac, great [20:44] bac, fix for bug 974617 didn't work: my alaysis must have been failty :-( [20:44] <_mup_> Bug #974617: test_operational_view_integration fails intermittently in parallel tests

< https://launchpad.net/bugs/974617 > [20:44] faulty [20:45] analysis [20:45] gary_poster: dang [20:45] my typing is faulty too [20:45] so is it only failing in paralleltests? [20:49] gary_poster: see you tomorrow. have a good evening [20:49] bye bac [20:49] you too