[12:27] bac benji danilos gmb, call in 3 [12:28] ack [12:28] ok [12:29] ack [12:44] gary_poster: what is the right disposition for this keyring card? [12:46] benji, um...needs info is good. that's what you said. [12:46] and just leave it where it is? === gary-lunch is now known as gary_poster [12:51] benji, oh on the kanban board...how about "move the bug number from the magic bug field to the title, and move it to Archive/Rubbish"? [12:51] sounds good [12:52] cool [12:56] hey benji, can you tell me whether you see messages from me in #launchpad-dev, to jelmer? [12:57] gary_poster: at 08:50 there's one where you mention his name [12:57] benji, cool, thanks [13:00] gary_poster, sqltrace is very nice, we need some of this in the OOPSes as well :) === gary-lunch is now known as gary_poster [13:02] * gary_poster appears to be having connection probs [13:25] bac benji danilos gmb, especially gmb because of your interest, anyone fancy updating ec2 test to have a new launchpad-dependencies deb? It's pretty easy, even with docs IIRC, and I'm happy to help. If not, I'll do it. [13:26] gary_poster: If it needs doing today, then I'll have to pass [13:26] gary_poster: i'd like to do it [13:26] bac, ok great. looking for instructions [13:26] if no one else is interested, I'll... I guess that was a hot property :) [13:27] lol [13:27] bac, https://dev.launchpad.net/EC2Test/Image [13:46] gary_poster, btw, would you be able to give me any hint for the following: https://pastebin.canonical.com/51923/? [13:46] danilos, looking [13:47] gary_poster, (other than uninstall system setuptools and ClientCookie :)) [13:49] danilos, yes. short answer, I think that you'll find your ClientCookie egg_info is obviously broken when you look at it. You might also find that it was supposed to have been ininstalled, but maybe not. Remove the egg_info manually, and then reinstall ClientCookie if desired. [13:49] long answer, there's a bug somewhere I filed...looking [13:50] bug 834698 danilos [13:50] <_mup_> Bug #834698: setuptools.egg-info can end up as a directory when it is meant to be a symlink < https://launchpad.net/bugs/834698 > [13:50] This is the ClientCookie version of the same problem [13:50] danilos, thank you for answering user story dude btw [13:51] gary_poster, oh, I have both setuptools (in the middle of the traceback, needs scrolling to the right) and ClientCookie messed up, weird [13:53] sadly not that weird. [13:53] Or at least not that unusual [13:55] gary_poster, but why did it happen all of a sudden? [13:56] gary_poster: in VALID_AMI_OWNERS is the name just any identifier? i expected it would be launchpad id but it clearly isn't. [13:57] danilos, I suspect that the problem has been there for a long time. Chameleon exposes it. Chameleon compiles templates to Python bytecode, so it wants to verify that its code it caches is up-to-date with your packages, so it wants to get the versions of all your installed packages to make a hash. Kinda grody, but not unreasonable AFAICT. [13:57] bac, your ec2 account #, I believe [13:57] gary_poster, ah, fair point then, thanks for explaining it all for me [13:57] welcome [13:58] gary_poster: yes, the key is the AWS account number. the value just seems to be a random name/nickname [13:59] bac, oh! lemme look [13:59] perhaps only used in logging? [14:00] the best part is, clientcookie is not packaged in ubuntu since lucid it seems [14:01] heh [14:01] bac, it looks there is a new "images" command that uses the mapping. I think it used to be a tuple before. It's just for reporting to humans, AFAICT [14:02] bac, lib/devscripts/ec2test/builtins.py is where I see it [14:02] bac, so I'd use "bac" [14:06] gary_poster: i planned to use 'bac', which clearly works for me as i'm mononicked, but thought i would document it. [14:06] bac, +1 [14:07] bac, do you feel comfortable with the answers I gave you, or would you like me dig more into something? [14:07] gary_poster: no that's fine. i was just double checking [14:08] cool bac. thanks for pursuing that. [14:24] gary_poster: when running bin/ec2 update-image what do you use for the base image? is it the ami identifier of the last one created? [14:25] bac, mm, good question. I think we use the base Lucid image provided by Canonical. Looking around... [14:26] bac, yeah, that's what it says on the wiki page I gave you--it even gives you the number to use and where to verify it [14:50] * gmb -> collecting car from the garage. Will be back in ~30mins, £250 shorter. Ugh. [15:04] gary_poster: based on http://uec-images.ubuntu.com/releases/10.04/ it looks like the image listed in the wiki page is old by several releases (2010-10-20 vs 2011-07-19). unfortunately i can't find an official page like http://aws.amazon.com/amis/4348 that describes it. i'm forging ahead with the newer ami as the base. [15:05] ok bac [16:01] gary_poster: I'm checking out. I'll see y'all tomorrow. [16:01] bye benji [16:19] gary_poster: so the new image is made and installed [16:19] it is now just automatically picked up via the new rev number? [16:20] great bac! So now you need to land the branch with your blessing, and then people can use it [16:20] bac, then maybe shoot an email to the list ccing stub letting folks know [16:20] gary_poster: no testing? just wait for a big boom [16:20] bac, heh, no, testing probably a good idea :-) [16:21] gary_poster: so, just do a run of ec2 test on devel? should i have done that before landing the branch that updates VALID_AMI_OWNERS? [16:24] ah, i see how it finds images [16:24] bac, do a run: yes. If it starts and tests are running it's probably fine [16:25] if you're already listed in VALID_AMI_OWNERS there is no way to test the image before subjecting it to the team [16:26] bac, there is [16:26] do not make the image public [16:26] oh, right [16:26] perhaps i'll update the wiki to mention that... [16:53] danilos, I think I need to ask you a question. If you are about to leave, gimme 5 if you can. One sec, getting details. :-) [16:54] danilos, heh, yeah [16:54] So, I just had to qa-bad my revision [16:54] that used your new getMergeProposals [16:55] the problem was that it worked great if I passed for_branches=[something] [16:55] but if for_branches=[]... [16:55] you can guess what happened :-P [16:55] So, obviously, I can change my call to not call getMergeProposals if I have no branches [16:56] but it struck me that it might be nice if getMergeProposals treated None differently than an empty iterable [17:08] So that an empty iterable would return an empty value [17:09] * bac -> food [17:16] gary_poster <- food [19:52] danilos, if you come back and look at this, I did change getMergeProposals as I described above: https://code.launchpad.net/~gary/launchpad/bug724025/+merge/73432 [20:32] hey gary_poster, my ec2 test of the new image is failing. :( [20:32] it is a twisted timeout test [20:33] i should be able to use the AWS web interface to change the image back to private and then try again [20:35] on call bac, make sure that it is not spurious and/or does not fail on current image [20:36] gary_poster: retesting just that test on new image [20:36] original run is not finished yet so i don't know the full set of failures [21:21] the test seems to be spurious on ec2. leaving new image in place. [21:21] good, bac