[00:14] <cjwatson> Anyone care if I upgrade dogfood?
[00:14] <cjwatson> May as well get my QA done.
[00:16] <cjwatson> https://lpbuildbot.canonical.com/waterfall looks a bit confused.
[00:16] <cjwatson> ("slave lost")
[00:22] <cjwatson> OK, upgrading mawson now.
[00:22] <wgrant> cjwatson: Go ahead.
[00:22]  * wgrant pokes buildbot.
[00:22] <wgrant> Ah, it was restarted.
[00:43] <cjwatson> mawson r14517 -> r14539.
[01:06] <cjwatson> Can anyone tell if there's something wrong with https://dogfood.launchpad.net/ubuntu/+source/libunicode-collate-perl/0.87-1/+build/2802211?  It seems to be taking ages to upload.  I don't see anything of interest in mawson:~launchpad/poppy.log.
[01:07] <cjwatson> I uploaded the source to dogfood so presumably poppy was at least somewhat working at some point.
[01:10] <cjwatson> Oh, maybe process-upload.py -C buildd
[01:10] <cjwatson> There it is.  Sorry to bother.
[01:31] <wgrant> cjwatson: That doesn't go via poppy.
[01:31] <wgrant> You can re-cron the process-upload for a while if you want.
[01:31] <wgrant> It doesn't run all the time because puppet.
[03:06] <wgrant> Hm
[03:06] <wgrant> Someone didn't run milestonetag through ec2.
[03:06]  * wgrant reverts.
[03:16] <lifeless>  thanks
[03:17] <wgrant> Actually, I'll just fix it.
[03:17] <wgrant> Missing UPDATE permissions for person merging.
[03:17] <lifeless> I'm not sure its the right answer to the problem
[03:17] <wgrant> Oh?
[03:17] <wgrant> It's not in a UNIQUE, so it doesn't need to be a 3-parter.
[03:17] <lifeless> tagging milestones, I mean
[03:17] <wgrant> The existing merge code in devel will handle it fine.
[03:18] <wgrant> Ah.
[03:18] <wgrant> So should I revert it by pretending the fix isn't trivial? :)
[03:18] <lifeless> well, I think you should be on leave, as I am
[03:18] <lifeless> :)
[03:18] <wgrant> True, true.
[03:20] <lifeless> I have an email thread to reply to about it
[03:20] <lifeless> I'm glad the squad moved forward
[03:20] <lifeless> ETRICKY
[03:21] <wgrant> It'll be on prod on Wed or Thu if you don't argue soon.
[03:21] <wgrant> Well.
[03:21] <wgrant> I guess that assumes that people remember to do fastdowntime.
[03:21] <wgrant> Which seems unlikely, but we'll see :)
[03:21] <lifeless> stuart will be on leave
[03:21] <lifeless> soon
[03:21] <lifeless> we have 2 FDT's to do before this could be a candidate
[03:21] <wgrant> Yes.
[03:22] <wgrant> Although I think we could really sensibly do the first one live.
[03:22] <lifeless> if it lands, and we decide it was wrong, its not a huge deal
[03:22] <wgrant> It's a very low-use table.
[03:22] <wgrant> True.
[03:22] <lifeless> wgrant: if anything references the table, it will get related PK row locks taken out on updates even if the table isn't mentioned in the query
[03:23] <wgrant> Sure.
[03:23] <lifeless> which table is it again?
[03:23] <wgrant> sourcepackagerecipedata
[03:23] <wgrant> Which should only be referenced by sourcepackagerecipedatainstruction.
[03:24] <wgrant> Oh, and build.
[03:24] <wgrant> Inconvenient.
[03:24] <lifeless> :)