[08:35] hello there... [08:35] has someone tested the alpha2 release here? === chrisccoulson_ is now known as chrisccoulson === bladernr_afk is now known as bladernr_ [14:51] TheYsNoi - lot's of people have. any problem? [14:54] brendand: he/she isn't here any more, [14:54] anyhow I suspect it's confusion between this channel and #ubuntu-testing, or something [15:03] hmm... rls-mgr-p-tracking seems to be missing some bugs that should be there. Wonder what glitched up now? [15:17] skaet, the glitch is "Friday, 20. January 2012 02:01 UTC " [15:17] last refresh is 2 weeks behind [15:18] *sigh* [15:18] thanks jibel [15:18] missed that [15:19] * skaet was focused on A2 since returning from vacation. [15:20] I complained about it in a release meeting or two. [15:20] So far I seem to have been ignored. [15:24] cjwatson, bjf is the person to ping directly when it stops working. Sorry I thought it had been taken care of, but didn't check for myself until this morning. [15:25] Somebody (Steve, I think) said they had already done so. [15:25] I've given up on that whole system as honestly it's just broken too often. At least LP is consistent. [15:25] Even if it's less well-categorised. [15:26] I wanted it in LP, but there were roadblocks put up for getting the categorization in there. [19:14] fyi, I need to take the publisher offline [19:15] we have an important firefox update that has to go out today, and I need to make sure everything that went to -updates with the rapid release migration makes it to -security [19:15] now that I know about sru-release, this should not be as painful as last time [19:16] (not nearly-- it should be off for less than an hour) [19:16] sorry, this should be the last of these [19:18] cron job disabled [20:19] jdstrand: I'd like to know how long it took [20:19] cjwatson: well, it is for two releases, maverick and lucid [20:19] lucid is done, maverick is still going [20:19] I didn't start sru-release right away though, cause there were a number of things I needed to copy-package [20:20] I think it was in the neighborhood of 25 minutes [20:20] and by they, I mean lucid [20:20] or it [20:20] meh [20:20] lucid I think took 25 minutes [20:21] which is *much* better than the 2.5+ hours it took for copy-package [20:24] OK. Archive.copyPackages might be faster yet, but that doesn't send mail so I want to think about whether it makes sense for sru-release to use that [20:24] How many packages was this? [20:26] * micahg was wondering about recent mail sent for -updates [20:30] Yeah, side-effect [20:31] But copyPackages might be a sensible solution actually [20:31] I forget what the typical overhead of an API request is [20:32] I'd like to get to the point where - at least - this sort of thing only requires good timing rather than stopping the publisher, and ideally is transactional so you always get either all or none [20:33] cjwatson: looks like 736, based on what q accept tells me I had to do with unapproved in lucid-security for language-pack- [20:33] (IIRC copyPackages isn't transactional though) [20:33] cjwatson: stuff was also coming from 2 different sources (-proposed and -updates) [20:34] fwiw, I commented out temporarily the copy to -updates in sru-release since they are already there [20:34] jdstrand: wow. one API request versus 736 might well make a serious difference then. [20:34] micahg: ok, two requests then :) [20:34] yeah, I was pretty surprised by that number before, and was resurprised today :) [20:34] I have to do it twice too [20:34] that might well just amount to good timing [20:34] ok maverick done [20:35] but both needed to be atomic, I guess the -proposed stuff could've been copied to -updates first, then copy everything to -security [20:35] strictly you only need to be confident of it completing before the next publisher run starts [20:35] if it takes five minutes or something, that's just timing [20:36] maverick had 770 [20:36] ok, almost done with my bits [20:38] of course it's a bit awkward that you have to accept from the queue as well [20:41] it is [20:43] ok, cron job re-enabled [20:45] if copyPackages were restricted to the archive owner (I don't *think* it is ...) or if we had an alternative privileged interface, then we could arrange for that to bypass the queue [20:46] it goes through the queue right now because we're using an interface that only requires ordinary uploader privilegeq [20:46] s/q$// [20:46] interestingly, it seems there might be some logic there [20:47] the langpacks like I said I didn't copy to -updates (commented out those lines in sru-release) [20:47] those needed to be approved [20:47] but when bzr reverted sru-release and used it for firefox and mozvoikko for lucid-oneiric, I was told to look in the q, but they weren't there [20:49] * jdstrand checks again [20:49] oh, there they are [20:49] must have been another cron job [20:50] ok, all accepted [21:16] yes, PackageCopyJobs are processed by a cron job [21:16] it's every minute or two I think === rsalveti` is now known as rsalveti