[10:42] frankban, Do you have any idea how I can ssh in to an ephemeral lxc instance on the buildbot slave? [10:43] user: ubuntu password: ubuntu doesn't work... [10:50] gmb: you have to use '/var/lib/buildbot/.ssh/launchpad_lxc_id_rsa' as ssh key (or something like that). you can use lp-lxc-ip -n lptests to get the container's ip [10:50] frankban, Ah, thanks. [10:53] gmb: remember to stop buildbot master while you are working on lptests lxc, otherwise you can generate a time/space fault... [10:53] Wow. [10:53] I'll bear that in mind :) [10:53] :-) [11:33] morning y'all [12:06] * bac wonders how this google hangout event will work. is the hangout pre-created for us? [12:08] bac, yes [12:08] I went to event page [12:08] then there is a link to hangout [12:09] bac benji frankban (gmb) https://plus.google.com/hangouts/_/b52c1a893347c571cd1a315088194860867fb396?authuser=0&hl=en-US in 2 [12:26] it is annoying that google events don't allow you to set up recurring ones...takes a lot of the utility out of it for a daily call. [12:39] agreed [12:43] hi gary_poster, YetAnotherDistributeTypeQuestion [12:44] bac, heh, ok [12:44] gary_poster: benji and i ran into this on our dev boxes and for expediency decided the simple thing to do was use sudo to go ahead and let it install in /usr/local/lib [12:44] http://paste.ubuntu.com/1064215/ [12:45] bac, I'd try sudo apt-get install python-setuptools first [12:45] i don't want to use 'sudo' in the verify_command for tarmac. is there another solution? or just use 'sudo' once manually and let it do the deed? [12:45] I think once it is installed, even the official package, things will work [12:45] and that's nicer [12:45] the official package installation is nicer I mean [12:47] oh, ok. just make a dependecy on python-distribute? [12:48] er, python-setuptools [12:48] which, unfortunately, is already installed [12:48] so installing python-setuptools does not help [12:49] ah :-/ [12:49] it worked for lpsetup :-/ [12:53] bac, question bout STDERR: and STDOUT: stuff in subunit from zope.testrunner... [12:53] Should it be: [12:53] STDERR: [text]\n [12:53] or [12:53] STDERR:\n [12:53] [text] [12:53] ? [12:54] gmb: the latter [12:54] ok. [12:55] bac, So, my question is, how would testr handle this: [12:55] successful: lp.codehosting.codeimport.tests.test_worker.TestSubversionImport.test_sync [ multipart [12:55] Content-Type: text/plain;charset=utf8 [12:55] STDERR: [12:55] 63 [12:55] No handlers could be found for logger "root" [12:55] WARNING:root:N changeset 2 [12:55] gary_poster: yeah, you see at line 11 of the paste that it sees the debian pkg is installed but is unhappy b/c it isn't an egg [12:55] WARNING:root:N changeset 3 [12:55] 0 [12:55] ] [12:55] Is that safe because it's in the multipart []? [12:56] gmb: i don't know what you mean by safe? it is just informational and should be ignored b/c it is in the multipart [12:56] bac, I mean "will it cause testr to choke and give us an unknown worker?" [12:56] gmb: i wouldn't think so [12:57] Okay. [12:57] Hmm. [12:57] bac, maybe is that an older ez-setup? [12:57] try the distribute-setup you have and see if it is more polite? [12:57] as I said, lpsetup worked with python-setuptools for me yesterday [12:58] gary_poster: you mean bootstrap.py? [12:58] gary_poster: yeah but lpsetup isn't using buildout [12:59] bac, no...I meant ez_setup. but yeah, I guess the part that is unhappy is bootstrap :-/ [12:59] right, so there is no ez_setup as bootstrap fetches it [12:59] does it fetch ez_setup or the distribute version? There is a --distribute falg now I think [12:59] flag [13:00] gary_poster: could we move our call up to 3pm today? [13:00] benji, could we move our call up to 2pm today? [13:01] gary_poster: sure [13:01] thanks [13:01] bac, sure :-) [13:01] gary_poster: it is this version: $Id: bootstrap.py 110538 2010-04-06 03:02:54Z tseaver $ [13:01] i don't see any option for using distribute. [13:01] if there is a newer bootstrap.py i could replace with it [13:02] bac, http://svn.zope.org/*checkout*/zc.buildout/trunk/bootstrap/bootstrap.py?rev=123006 [13:02] that has a --distribute [13:04] bac, oh one other idea. if that falls over, try this one: [13:05] bac http://svn.zope.org/*checkout*/zc.buildout/branches/1.4/bootstrap/bootstrap.py?rev=116869 [13:07] gary_poster: i tried with the first new boostrap.py you referenced. with it 'python bootstrap.py' worked, no new option needed [13:07] bac, great, hopefully [13:10] it is ashame that those bootstrap files often don't have any date or versioning information embedded within them [13:17] that's one of the reasons benji probably advocates not including bootstrap in packages, and forcing people to download them. I disagree, but I see where he is coming from. :-) [13:18] gary_poster: well, i'd like you to review this instead of benji: https://code.launchpad.net/~bac/zope.testing/newbootstrap/+merge/112562 [13:19] :-) [13:19] gary_poster: i did not bump the version as it does not need releasing [13:20] bac, ack. I made an "approve" comment but did not approve the mp (should I, as a matter of best practice, do you think?) [13:21] gary_poster: if it does not require any code changes then i think the reviewer should do it. that prevents any subsequent versions from getting merged...which has good and bad points to it. [13:21] bac, cool. approved mp [13:24] gary_poster: well this is odd: [13:24] 2012-06-28 13:24:03 INFO No approved proposals found for lp:~launchpad/zope.testing/3.9.4-fork [13:26] frankban, I was just reviewing the results log. bug 974617 showed up on 2012-05-15 . then it was completely silent till 2012-06-21, when it had two instances--and after which we've had two more instances in the next seven days. Do you remember anything off-hand that might have re-triggered the problem? Did we reduce the timeout again? [13:26] <_mup_> Bug #974617: test_operationalerror_view_integration fails intermittently in parallel tests

< https://launchpad.net/bugs/974617 > [13:27] ISTR we had some very large timeout [13:27] and then we reduced it to 60 [13:27] and then increased it to 180 [13:27] (we reduced it to 60 when you improved the retry code somehow) [13:27] bac, uh? [13:28] bac, I'm afraid I don't know what to do about it. [13:28] do you have any thoughts? [13:28] gary_poster: was just informational [13:28] oh cool [13:28] gary_poster: i'm chatting with dobey [13:28] cool [13:31] gary_poster: the connection is now retried for 180 times [13:32] gary_poster: despite the vague message, the problem was the MP lacked a commit msg so it was ignored. boo. [13:33] bac, bah [13:33] frankban: https://bugs.launchpad.net/lpsetup/+bug/1016645 calls for positional arguments for repo and branch; is there an easy way to do that in the current lpsetup structure? [13:33] <_mup_> Bug #1016645: lpsetup: get command < https://launchpad.net/bugs/1016645 > [13:33] I might be better off using switches instead. [13:34] bac, a vague message that you could only see if you logged on to the tarmac machine, no less [13:34] frankban, https://plus.google.com/hangouts/_/777c438088f7788065d36f022db674481b12d167?authuser=1&hl=en-US when you are ready [13:35] benji: you can add positional arguments as you usually do with argparse [13:35] but now when people are confused there are two of us to ask "did you set a commit message"? [13:35] benji, argparse supports that, and we're on top of that; don't have a ready answer to your question, but am hopeful that there is one. That said...there's probably reasonable defaults for those values, which suggest switches, and it seems to be the path of least resistance anyway, so +1 [13:36] bac, heh, true [13:36] gary_poster: it probably wouldn't be hard, but switches make more sense to me anyway [13:36] cool [13:38] benji: however: http://stackoverflow.com/questions/4480075/argparse-optional-positional-arguments [13:40] frankban: thanks [14:13] gary_poster, Do you have some time to help me debug our parallel testing 2.7 problems? I've been having issues getting into the LXC containers on the slave in order to be able to follow the steps you outlined in your email. [14:15] gmb, on call, wil ping [14:15] Sure [14:20] frankban: am I right in thinking that "repository" referenced in https://bugs.launchpad.net/lpsetup/+bug/1016645 is the same as the already-existing --directory option? [14:20] <_mup_> Bug #1016645: lpsetup: get command < https://launchpad.net/bugs/1016645 > [14:45] benji: looking [14:45] gmb, hey sure. how would you like to pair? [14:45] gary_poster, Sure thing. [14:45] I'll start a hangout. [14:45] cool [14:45] gmb, I will "prepare." Back in 2. [14:46] O.o [14:46] benji: yes, it's the same [14:47] frankban: cool; I got half-way into implementing it when I realized it was already done but with a different name. [14:47] :-( [14:48] no worries [14:51] gmb, hangout? [14:51] gary_poster, G+ is teh suck. [14:51] Bear with me... [14:51] :-/ k [14:53] gary_poster, https://plus.google.com/hangouts/_/713717cf5af4c438fee987556699093cd56fd288?authuser=0&hl=en-GB# [14:54] gary_poster: the next step for tarmac-controlled branches would be to move them to be owned by lpqabot so that people cannot just push to them directly as currently done with the launchpad branches and the PQM user. for the two under our control now i don't think that's a high priority but should be done before we hand it off to lp-proper. agree? [14:55] agree bac [15:26] I have an MP ready for review: https://code.launchpad.net/~benji/lpsetup/bug-1016645-add-command-line-options/+merge/112588 [16:06] benji, I'll review after lunch if no-one beats me to it. I request/suggest a low priority card for branch option. Also, if I were reviewing, I'd ask your opinion on the spelling of directory vs repository. I'd have to think through it myself, and asking you would help. :-) [16:10] gary_poster: I don't love "directory" but not being steeped in bzr-lore, "repository" doesn't seem to mean the right thing either; at least "directory" is meaningless enough that it can mean whatever we want it to mean given the context [16:22] gary_poster: do you thing initlxc should stop the container at the end of the process? [16:22] s/thing/think [17:02] frankban, good question. thinking... [17:03] frankban, to make our normal story as smooth as possible, the answer would be, by default, no. Maybe an option to turn it off at the end would be good. [17:04] Leaving it on make me kinda nervous, but I think it makes the most sense. Telling the user that the container is running woud be a good thing to do [17:04] gary_poster: cool, that's what I was thinking. [17:04] cool [17:08] benji, repository vs. directory: aren't we talking about the argument to initrepo? [17:09] stepping away, back soon [17:09] gary_poster: I don't know about initrepo, but it's an argument to just about all of the commands because it's where all the files go [17:10] Now I'm doubting my understanding of the situation... but frankban said I was right... [17:12] benji: I think gary_poster meant "bzr init-repo" and yes, directory is the argument init-repo in this case. so, it's actually a bzr shared repository. +1 on renaming directory to repository [17:28] benji, what frankban said :-) [17:29] gary_poster: k [18:26] gary_poster: the changes to rename "directory" to "repository" are done [18:26] benji, thank you. I've had some other things come up, sorry [18:27] no worries [18:37] benji https://plus.google.com/hangouts/_/096fce3ced6cd2c17da53177decc22d9ae8a5b43?authuser=1&hl=en-US ? [18:37] when you are ready [18:37] no rush [19:09] gary_poster: ping [19:09] bac, I know, five more mins max, sorry [19:10] gary_poster: ok [19:18] bac https://plus.google.com/hangouts/_/096fce3ced6cd2c17da53177decc22d9ae8a5b43?authuser=1&hl=en-US [19:37] * benji enjoys a mango lassi. [21:42] * bac braces for leankit-pocalypse tomorrow [23:36] gmb you must share with frankban tomorrow. i concur. https://sphotos.xx.fbcdn.net/hphotos-ash4/s720x720/293727_10150902819136845_208774280_n.jpg