| infinity | tjaalton: I assume that xserver-autobind-hotplug.patch you dropped in your xorg-server SRU was unused/unapplied? | 07:43 |
|---|---|---|
| infinity | Guessing so, given the lack of change in series. | 07:43 |
| infinity | tjaalton: Would have been nice to mention it in the changelog, so I didn't have to guess. ;) | 07:44 |
| infinity | Oh. That patch wasn't even in debian/patches, it was just cruft in the root. | 07:46 |
| infinity | Derp. | 07:46 |
| LocutusOfBorg | slangasek, I'll answer soon wrt virtualbox meh :) | 07:46 |
| LocutusOfBorg | do you have any ETA? | 07:46 |
| ogra_ | infinity, http://paste.ubuntu.com/20424372/ ... *sniff* | 08:09 |
| ogra_ | running "sudo snapcraft" locally works though ... but cleanbuild falls over | 08:10 |
| ogra_ | cjwatson, does LP actually run snapcraft cleanbuild when it builds snaps ? | 08:11 |
| infinity | ogra_: What is "cleanbuild"? | 08:11 |
| infinity | ogra_: Forces it into a container with restricted mounts and ickiness? | 08:11 |
| ogra_ | it makes snapcraft run the whole build inside an lxc container | 08:11 |
| infinity | ogra_: That would be unnecessary in LP. | 08:11 |
| infinity | ogra_: Since LP builds each happen in a fresh environment anyway. | 08:12 |
| ogra_ | well, does the build actually run as root ? | 08:12 |
| ogra_ | i seem to actually get a proper build when i run it locally as root | 08:12 |
| ogra_ | (it is just snapping my build result) | 08:13 |
| ogra_ | taging livebuild | 08:13 |
| ogra_ | Priming livebuild | 08:13 |
| ogra_ | Snapping 'ubuntu-core' | | 08:13 |
| ogra_ | Snapped ubuntu-core_16.04.1_amd64.snap | 08:13 |
| ogra_ | ogra@styx:~/Devel/images/snappy/build$ | 08:13 |
| ogra_ | :D | 08:13 |
| ogra_ | so the question is how an LP snap build is actually invokend i think | 08:14 |
| infinity | Should be easily determined by looking at a build log. | 08:16 |
| infinity | Or by reading the launchpad-buildd source. | 08:17 |
| ogra_ | hmmwell, trying to fins a snap buildlog :) | 08:17 |
| ogra_ | *find | 08:18 |
| * ogra_ notes that the form to set up a snap build defaults to yakkety ... probably not what we want | 08:18 | |
| ogra_ | well ... why bother ... i'll just try it ... | 08:32 |
| ogra_ | https://code.launchpad.net/~ogra/+snap/os-snap-test/+build/1743 | 08:32 |
| infinity | ogra_: Looks happy. | 08:41 |
| ogra_ | OOOOOHHH !!! | 08:41 |
| ogra_ | https://launchpadlibrarian.net/274437775/buildlog_snap_ubuntu_xenial_amd64_os-snap-test_BUILDING.txt.gz | 08:41 |
| ogra_ | yeah ! | 08:41 |
| ogra_ | except that i now have a non-os snap that contains an os nap and two kernel snaps :) | 08:41 |
| infinity | ogra_: And I note the point release version there, too. You fixed it to grub around in os-release? | 08:41 |
| infinity | ogra_: Can one build produce multiple snaps, or will this be a Complicate Problem to solve? | 08:42 |
| ogra_ | there seems to be something worng with using the PPA in live-build ... the inner snaps are actually anonical-pc-linux_4.4.0-31+20160722.08-39_amd64.snap and ubuntu-core_16.04+20160722.08-37_amd64.snap | 08:42 |
| ogra_ | i think you can only produce one snap | 08:43 |
| infinity | That seems short-sighted. | 08:43 |
| ogra_ | there are definitely still some changes ahead, but the prerequisites are there \o/ | 08:43 |
| infinity | Especially with the content-sharing stuff, it would make sense to want to produce N snaps with content-sharing deps. | 08:43 |
| infinity | From one build. | 08:43 |
| ogra_ | i can produce snaps for multiple arches | 08:43 |
| ogra_ | with the same name | 08:44 |
| ogra_ | but i cant produce kernel and os snaps with different names from one build | 08:44 |
| ogra_ | (which is fine) | 08:44 |
| ogra_ | i roughly know what to do now ... | 08:44 |
| infinity | Good deal. | 08:45 |
| infinity | Nice when a plan comes together. | 08:45 |
| ogra_ | also snapcraft will need to learn to mangle the yaml... what it currently produces acnt be used as rootfs | 08:45 |
| ogra_ | well, i guess i need signoff from mark still ... he needs to understand that the machinery is identical to what produces the cloud images and that i only replace the wrapper | 08:46 |
| ogra_ | i dont want to be shouted at *after* i implement the changes :) | 08:46 |
| ogra_ | infinity, hmm, | 08:56 |
| ogra_ | Get:16 http://ftpmaster.internal/ubuntu xenial-updates/main amd64 livecd-rootfs amd64 2.408.2 [48.1 kB] | 08:56 |
| ogra_ | i dont get why it pulls from xenial-updates | 08:57 |
| ogra_ | https://launchpad.net/~snappy-dev/+archive/ubuntu/image/+packages?field.name_filter=&field.status_filter=published&field.series_filter=xenial ... | 08:57 |
| ogra_ | livecd-rootfs - 2.420+ppa15 | 08:57 |
| ogra_ | it should pull that version instead | 08:57 |
| infinity | ogra_: Only if it's building in a PPA... | 08:58 |
| ogra_ | ah, so i picked the wrong archive at the form ... | 08:58 |
| * ogra_ tries changing that | 08:58 | |
| infinity | ogra_: Yeah, I see snappy-dev/tools there, but you wanted snappy-dev/image, right? | 08:59 |
| ogra_ | yep | 08:59 |
| ogra_ | well ... | 09:00 |
| ogra_ | EXTRA_PPAS='snappy-dev/image snappy-dev/edge' | 09:00 |
| ogra_ | that is what it shoudl use | 09:00 |
| infinity | Aaand, that's where things might fall over. | 09:00 |
| infinity | The EXTRA_PPAS thing might be specific to livefs builds. | 09:00 |
| infinity | Though, I'd think that PPAs with deps on other PPAs would behave similarly. | 09:01 |
| infinity | Yeah, EXTRA_PPAS is a livefs-only option, other types of LP builds can't specify anything other than archive. | 09:01 |
| infinity | That said, if a PPA has a dep on another PPA, both should end up in sources.list. | 09:02 |
| infinity | So, perhaps you need a hierarchy like edge -dep-> stable -dep-> image | 09:02 |
| infinity | (Fake PPA names, I don't know what you have) | 09:03 |
| infinity | But you get the idea. | 09:03 |
| ogra_ | well, i see the ppa in the apt-get update output of the build | 09:04 |
| infinity | Just the one PPA that you're building in though. | 09:04 |
| xnox | infinity, i'm pretty sure that one has to manually add recursive ppa dependencies by hand. Only direct PPA deps end up in sources.list. | 09:05 |
| ogra_ | i just triggereed a build with the snappy-dev/image PPA as source ... lets see | 09:05 |
| ogra_ | the prob will be that we use two PPAs currently | 09:05 |
| infinity | ogra_: Well, what's in edge? Stuff needed to build, or stuff needed in the snap? | 09:05 |
| ogra_ | and LP only allows me to pick one | 09:05 |
| ogra_ | edge only has the daily snapd build after it ran though elopios tests | 09:05 |
| xnox | ppaA (which build deps on ppaB), and if your ppa you want both, one has to specify both | 09:06 |
| infinity | ogra_: What you probably want is "stuff neeeded to build" -> image PPA, and then "stuff needed in the chroot created by live-build" -> specify in build. | 09:06 |
| ogra_ | i see live-build pull updates for both PPAs in the build ... that looks more like some kind of pinning is going on based on what i chose in the LP form | 09:06 |
| infinity | Anyhow, let's see your results and see if something needs tweaking. | 09:07 |
| ogra_ | yeah | 09:07 |
| ogra_ | this is incredibly fast btw ... i wonder why ... | 09:07 |
| ogra_ | the last build only took 7min | 09:07 |
| infinity | Should it not be? | 09:08 |
| infinity | It's a tiny rootfs. | 09:08 |
| ogra_ | iirc the livefs bulder takes more like 10-15 | 09:08 |
| infinity | It's the same builder. :P | 09:08 |
| infinity | And the same code. | 09:08 |
| ogra_ | for an amd64 build | 09:08 |
| ogra_ | hmm, k | 09:08 |
| ogra_ | then i'm probably wrong | 09:08 |
| infinity | So, you might just be hitting different scalingstack hosts. | 09:08 |
| infinity | Underlying hardware in scalingstack differs wildly. | 09:09 |
| ogra_ | ah | 09:09 |
| infinity | The machines there span about 4 years, AMD and Intel, crap and slightly less crap disk, etc. | 09:09 |
| ogra_ | aha | 09:09 |
| ogra_ | yeah, that will likely be it then | 09:09 |
| infinity | The joy of "cloud". | 09:09 |
| ogra_ | heh | 09:10 |
| DJones | HI folks, does http://changelogs.ubuntu.com/meta-release need updating now that 16.04.1 has been released so that LTS users will see 16.04.1 | 09:10 |
| infinity | You'll note that ppc64el and arm64 scalingstack are much more consistent, cause the compute nodes are all identical. | 09:10 |
| infinity | DJones: Not yet. | 09:10 |
| infinity | DJones: It'll happen next week after we've double-checked all the upgrade bugs, etc. | 09:10 |
| infinity | DJones: There's a reason the release announcement said that 14.04 users would be automatically upgraded "soon", rather than "now". | 09:11 |
| DJones | infinity: Thanks, just getting people asking in #ubuntu why they can't upgrade yet, noprobs,just wanted to make sure | 09:11 |
| ogra_ | infinity, ok ... | 09:13 |
| ogra_ | Snapping 'canonical-pc-linux' ... | 09:13 |
| ogra_ | Snapped canonical-pc-linux_4.4.0-31_amd64.snap | 09:13 |
| infinity | ogra_: Oh, the other major difference is often just network traffic and if a node is in the same DC as builddmaster, etc. | 09:13 |
| ogra_ | that definitely uses the right livecd-rootfs | 09:13 |
| infinity | ogra_: Because LP counts the "return result blobs to the master" in the total build time. | 09:14 |
| ogra_ | but likely not the snapd from the edge PPA :/ | 09:14 |
| ogra_ | ah | 09:14 |
| ogra_ | Get:19 http://ppa.launchpad.net/snappy-dev/edge/ubuntu xenial/main amd64 snapd amd64 2.0.10+ppa167-1 [4572 kB] | 09:14 |
| ogra_ | looks good | 09:14 |
| ogra_ | \o/ | 09:14 |
| infinity | Curious. | 09:14 |
| infinity | You must be specifying that in your live-build setup. | 09:15 |
| ogra_ | yeah, but as long as it works :) | 09:15 |
| ogra_ | i do | 09:15 |
| ogra_ | might be my Makefile hackery | 09:15 |
| ogra_ | (i'm not that good with shell scripts in makefiles :P ) | 09:15 |
| infinity | Heh. | 09:15 |
| infinity | Are you in the big hack room? | 09:15 |
| infinity | Or a session? | 09:15 |
| infinity | ogra_: ^ | 09:16 |
| ogra_ | http://bazaar.launchpad.net/~ogra/+junk/os-snap-test/view/head:/Makefile | 09:16 |
| ogra_ | big room | 09:16 |
| ogra_ | i guess thats just some quoting issue | 09:16 |
| infinity | Kay. Smoke break? Then I'll follow you back there. | 09:16 |
| ogra_ | yep | 09:16 |
| apw | ogra_, your variable assigments at the top are very odd, using variable expansion on the right ? | 09:18 |
| LocutusOfBorg | will anybody please accept libcypher-parser in queue? one rdep that I'll retry once it is published, a little soname change | 09:38 |
| cjwatson | ogra_: no, LP does not use cleanbuild. Anyway, as infinity says, you can just read lp:launchpad-buildd - "buildsnap" there does most of the work and it's very short. | 09:46 |
| ogra_ | cjwatson, yeah figured it all out and it works wonderfully https://code.launchpad.net/~ogra/+snap/os-snap-test/+build/1745 :) | 09:50 |
| ogra_ | (still needs some cleanup in livecd-rootfs ... but working snap builds actually mean i can drop 80% of the awful hacks :) ) | 09:51 |
| infinity | (Meaning 20% still remain) | 09:51 |
| ogra_ | :P | 09:56 |
| cjwatson | ogra_: https://code.launchpad.net/~ogra/+snap/os-snap-test/+build/1745 looks slightly implausible ... | 10:09 |
| cjwatson | ogra_: launchpad-buildd doesn't pick up livecd.* when doing snap builds | 10:09 |
| ogra_ | what looks implausible there ? | 10:09 |
| cjwatson | one 4K snap file? | 10:09 |
| ogra_ | yeah, but we wont need that at all anymore | 10:09 |
| cjwatson | I doubt that Ubuntu Core fits in 4K :) | 10:10 |
| infinity | Yeah, it still needs work. ;) | 10:10 |
| infinity | This is a WIP. | 10:10 |
| ogra_ | cjwatson, that might be because the install target in the makefile just calls "echo foo" atm ;) | 10:10 |
| infinity | The snap is intentionally empty currently. | 10:10 |
| cjwatson | Ah, OK. | 10:10 |
| ogra_ | i need to copy the rootfs chroot content in the installl target | 10:11 |
| ogra_ | and rip out all the internal snap creation bits | 10:11 |
| ogra_ | the important point is that this can auto-uppload to the store ... | 10:11 |
| ogra_ | compared to cdimage hackery thats a huge win | 10:11 |
| cjwatson | Yep | 10:12 |
| ogra_ | the big piece ahead is still to convince mark that this is better ... | 10:15 |
| tjaalton | infinity: yeah xenial had cruft in the diff :/ | 11:08 |
| infinity | tjaalton: Yeah, I sorted that out myself and accepted. | 11:17 |
| tjaalton | thx | 11:18 |
| tjaalton | infinity: are trusty dailies now with lts-xenial? | 11:38 |
| infinity | tjaalton: Not yet. Getting there. | 11:38 |
| tjaalton | k | 11:39 |
| infinity | tjaalton: They definitely should be on/by Monday, but maybe earlier. | 11:39 |
| tjaalton | infinity: going through NEW? mesa has been sitting there a few days, adds mesa-vulkan-drivers | 11:46 |
| tjaalton | and ppc64el ftbfs, can drop llvm again if it makes a difference | 11:46 |
| infinity | tjaalton: Yeah, I'm slowly going through it. | 11:47 |
| tjaalton | coolio | 11:54 |
| ogra_ | cjwatson, now the snaps have the right size ;) | 12:28 |
| cjwatson | cool | 12:34 |
| ogra_ | cjwatson, hmm ... do you filter out "type: os" in LP builds ? | 13:16 |
| ogra_ | if i put it in my snapcraft.yaml it stays around when i do a local build ... but i cant find it in meta/snap.yaml inside a LP produced snap | 13:17 |
| ogra_ | hmm, might actually be a livecd-rootfs issue after all (silly hacks) | 13:26 |
| cjwatson | ogra_: we do no such filtering | 13:38 |
| ogra_ | yeah, just found the issue | 13:38 |
| ogra_ | a very bad hack in livecd-rootfs plus a bug in snapcraft that it doesnt seem to fully replace meta/snap.yaml in the prime step | 13:38 |
| * ogra_ filed bug 1605622 | 13:42 | |
| ubot5 | bug 1605622 in Snapcraft "if something creates meta/snap.yaml during a snapcraft build, "type: os" is not carried over from snapcraft.yaml during prime step" [Undecided,New] https://launchpad.net/bugs/1605622 | 13:42 |
| infinity | ogra_: So, all sorted with your rm? | 13:45 |
| ogra_ | yep | 13:45 |
| ogra_ | all fine now | 13:45 |
| ogra_ | but there is something wrong in snapcraft indeed ... it should just overwrite the exiting file | 13:46 |
| ogra_ | the rm does fine for now though | 13:46 |
| bdmurray | slangasek: Its been brought to my attention that http://archive.ubuntu.com/ubuntu/dists/xenial-updates/main/dist-upgrader-all/current/ReleaseAnnouncement doesn't have the words LTS in it. Does that seem worth an SRU? | 15:28 |
| bdmurray | oh, he's out today. infinity -^ | 15:30 |
| === pleia2_ is now known as pleia2 | ||
| bdmurray | well, having the mirror lists updated too is probably a good idea | 16:01 |
| === pesari_ is now known as pesari | ||
| === debfx_ is now known as debfx | ||
| slangasek | bdmurray: ReleaseAnnouncement SRU> I checked and the trusty one doesn't say 'LTS' either... your call? | 17:18 |
| bdmurray | slangasek: the precise one does and I uploaded Xenial already | 17:19 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!