[11:41] <ogra_> would be nice if the two livecd-rootfs uploads could be approved before 5am UTC so the next auto-build of snappy wily doesnt fail
[12:15] <ogra_> slangasek, FYI i had to divert grub-install during rootfs build for snappy after adding the -signed packages
[12:23] <infinity> ogra_: And why is that uniquely required for snappy?
[12:23] <ogra_> dunno, why is grub-install run if i seed these packages ? it wasnt before it seems
[12:24] <infinity> Adding signed wouldn't change anything.
[12:25] <ogra_> no, it was a bigger change than that ...
[12:25] <infinity> Well, you implied it was just that. :P
[12:25] <ogra_> http://paste.ubuntu.com/12722981/
[12:26] <ogra_> seems the postinists of these packages behave differently
[12:27] <infinity> Well, you didn't have grub2 installed before.
[12:27] <infinity> So, yeah.
[12:27] <infinity> Quite different.
[12:27] <infinity> Erm, nevermind.  Can't read.
[12:27] <infinity> You didn't have grub2-efi installed before.
[12:28] <ogra_> right
[12:28] <ogra_> and we dont use the signed stuff yet but need it in the rootfs for ubuntu-device-flash consumption
[12:29] <infinity> The grub-efi postinst isn't meant to fail when grub-install fails, mind you.
[12:29] <ogra_> it complains about no such dir ...
[12:30] <ogra_> might not have been the failure that the build choken on ... there were two postinsts failing
[12:30] <ogra_> but both had the grub-install call
[12:30] <ogra_> *choked
[12:31] <infinity> The following packages have unmet dependencies:
[12:31] <infinity>  grub-efi-amd64 : Conflicts: grub-pc but 2.02~beta2-28 is to be installed
[12:31] <infinity>  grub-pc : Conflicts: grub-efi-amd64 but 2.02~beta2-28 is to be installed
[12:31] <infinity> That looks more problematic.
[12:31] <ogra_> that is why i seeded grub-pc-bin ;)
[12:32] <infinity> I don't actually see the failure you're referencing, though.
[12:32] <ogra_> i already did a test build with the same change on vivid ...
[12:32] <infinity> Do you have a pointer?
[12:32] <ogra_> https://launchpadlibrarian.net/220705208/buildlog_ubuntu_vivid_amd64_ubuntu-core-system-image_BUILDING.txt.gz
[12:32] <ogra_> or https://launchpadlibrarian.net/220705134/buildlog_ubuntu_wily_amd64_ubuntu-core-system-image_BUILDING.txt.gz
[12:32] <ogra_> if you want wily
[12:33] <infinity> That wily one is where I pasted from. :P
[12:33] <ogra_> and https://launchpad.net/~ubuntu-cdimage/+livefs/ubuntu/vivid/ubuntu-core-system-image/+build/40107/+files/buildlog_ubuntu_vivid_amd64_ubuntu-core-system-image_BUILDING.txt.gz passes after the change
[12:33] <ogra_> oh
[12:33] <ogra_> thats weird then
[12:33] <infinity> Oh.
[12:33] <infinity> grub-install: error: cannot find EFI directory.
[12:33]  * ogra_ re-reads
[12:33] <infinity> ^-- That?
[12:33] <infinity> Maybe you should have an EFI directory?
[12:33] <infinity> Instead of being hackish.
[12:33] <ogra_> that and a few lines down there is another one
[12:34] <ogra_> i really dont think we want to run grub-install at all ... i could be wrong though
[12:34] <ogra_> we just need the binaries in the rootfs so u-d-f can pull them when it creates the actual image
[12:35] <infinity> Possibly not.
[12:35]  * infinity shrugs.
[12:35] <ogra_> (our actual grub.cfg is also only 30 lines or so, quite different to a normal installation)
[12:36] <infinity> Err, really?  You don't use update-grub?
[12:36] <ogra_> (and snappy doesnt use update-grub either)
[12:36] <infinity> Because why would someone want something that was familiar? :P
[12:37] <ogra_> well, it messes up the rollback function ... snappy only needs to flick two vars back and forth, the rest is hardcoded anyway
[12:37] <ogra_> http://paste.ubuntu.com/12723088/
[12:38] <ogra_> the advamtage of creating complete disk images with hardcoded partitioning scheme ;)
[12:39] <ogra_> infinity, thanks
[12:39]  * ogra_ hopes it actually works without changed to u-d-f now :P 
[12:39] <ogra_> *changes
[15:13] <rcj> infinity, Could I have your help with an SRU in pending? bug #1473533
[15:14] <rcj> infinity, it's been 6 or 7 days which is barely time but it's affecting the azure cloud streams and causing a juju breakage now.
[15:15] <rcj> I was hoping that at least the precise package could be promoted to -updates today
[15:15] <rcj> Scope of change is very small and risk is as well
[15:18] <rcj> stgraber, arges, slangasek ^ Could one of you look at this SRU?  I just saw that infinity is under the weather.
[15:23] <arges> rcj: looking
[15:23] <rcj> arges, thanks. Not sure who can do what.  You were a sponsor on this upload
[15:23] <arges> rcj: so typically we don't release on Friday's in case of regressions and people not being around during the weekend etc
[15:31] <rcj> arges, I can understand that.  For me this is causing a Juju deployment breakage due to bad simplestreams data.  The fix is very narrow.  Without the fix the two functions will always throw an exception and do no meaningful work.  With the fix they will start returning a list of countries or timezones again.  But let me check that we can get the package applied in production if it were to be promoted.
[15:31] <arges> rcj: ack
[15:41] <ogra_> infinity, bah, crap ... there is another issue with hardcoded grub-pc in livecd-rootfs (i should have litened to you earlier :P )
[15:41] <ogra_> *listened
[15:52] <ogra_> infinity, ^^ that one actually fixes the conflict (i thoougth the seed change would have been enough, didnt know there was more hardcoding for grub-pc)
[16:01] <slangasek> ogra_: hrm, ok, if grub-install wasn't previously being called from the rootfs build, I can see that being required
[16:01] <slangasek> arges: are you releasing the SRUs for python-tz or still thinking about it?
[16:09] <arges> slangasek: I can do it, waiting on rcj. is it alright if its a critical fix
[16:09] <slangasek> it's certainly alright.  what is it you're waiting for?
[16:10] <arges> slangasek: i'll just release, I know typically we dont release on fridays for regressions but this seems like a pretty cut and dried case
[16:10] <rcj> arges, slangasek: thank you.  I had told arges that it could wait if I couldn't get it applied in production today.
[16:11] <rcj> but it would help to have it in -updates
[16:11] <slangasek> ah
[16:12] <slangasek> yes, this is a one-liner change that's backwards-compatible.  I think it should just be released
[16:12] <arges> ok doing it now
[16:12] <slangasek> and I'll be around email this weekend, so if there are regressions and someone files a regression-update bug I'll see it
[18:49] <smoser> anyone able to accept the cloud-init above for me?