[08:14] <ogra_> stgraber, did you eventually find the issue for the broken diff generation of system-image ?
[08:15] <ogra_> (would be really nice to be able promote image #6 on monday)
[14:47] <cjwatson> I've removed arm64 from OUTOFSYNC_ARCHES in proposed-migration.
[14:48] <cjwatson> So it'll now require arm64 builds to be up-to-date if the previous version was built.
[14:49] <cjwatson> It's caught up enough now that this should be safe.
[14:50] <apw> cjwatson, cool
[15:45] <cjwatson> apw: Are we getting an arm64 kernel package soon? :)
[16:49] <stgraber> ogra_: I didn't have time to investigate on Friday, will try to fix this today
[16:58] <apw> cjwatson, sure are ... waiting on some testing but yeah
[17:43] <ogra_> stgraber, well, i meant "on monday" ... i didnt mean "please work on the weekend" ;)
[20:11] <micahg> infinity: any chance of fixing the other arm64 buildds? 3 are on manual
[20:40] <apw> micahg, as likely as anything with the queue so short 'not birch' are disabled for performance, that one is much more reliable
[20:41] <micahg> apw: was unaware that they weren't all the same, makes sense
[20:41] <apw> micahg, birch is definatly superior
[22:12] <cjwatson> micahg: I believe we're deliberately sending failures to birch to see if it does better
[22:13] <cjwatson> (So much as apw says)
[22:13] <micahg> well, my build was fine the second time around
[22:14] <cjwatson> Yep, that's not too surprising
[22:15] <cjwatson> I expect we'll unmanual before EOD
[22:16] <cjwatson> i.e. before the next flood
[22:16] <micahg> ok, I've been seeing some failures on arm64 for previous uploads, I guess the only sane thing to do at the moment is fix if it's an autotools thing and leave it for someone else who has hardware otherwise?
[22:17] <micahg> or are the qemu instructions somewhere?
[22:17] <cjwatson> We don't have full-system qemu yet AFAIK
[22:18] <cjwatson> And I don't think the user-mode qemu is packaged
[22:18] <cjwatson> So indeed, fix if obvious, leave if not
[22:18] <cjwatson> It shouldn't be a blocker for anyone - arm64 failures will only block migration if it built before
[22:19] <micahg> that was the case I had :)
[22:19] <micahg> was some weird failure in dpkg-genchanges
[22:19] <cjwatson> Just retry that
[22:19] <micahg> yeah ,already did and it built fien
[22:19] <cjwatson> That's the non-birch hw being randomly segfaulty
[22:21] <cjwatson> We have a special version of launchpad-buildd installed on those builders that checks dmesg for the faults and considers them failures
[22:21] <cjwatson> At least I think that's still true
[22:23]  * cjwatson fails to understand why haskell-{sdl,cairo} and the stacks above them are showing up in http://people.canonical.com/~ubuntu-archive/transitions/ghc.html
[22:24] <cjwatson> They look installable to me and they aren't showing up in p-m output
[22:26] <cjwatson> I wonder if it's an :any dependency somewhere, doom
[22:26] <cjwatson> Oh well, nap time first