| wgrant | ubuntu-touch wily armhf builds are killing buildds again. http://paste.ubuntu.com/11363499/ | 04:35 |
|---|---|---|
| wgrant | infinity: ^^ | 04:35 |
| infinity | wgrant: Is that killing the host somehow, not just making the build explode? | 04:39 |
| wgrant | infinity: Host stops responding to ICMP. | 04:40 |
| wgrant | (and everything else) | 04:40 |
| infinity | Impressive. | 04:40 |
| infinity | wgrant: Guessing it's a live-build bug in that it's not correctly (or at all) cleaning up virtual FS mounts on error/exit. | 04:43 |
| infinity | wgrant: But not loving that that can crash a host. | 04:44 |
| infinity | wgrant: I guess we could perhaps blame buildlivefs instead, and say it should try a umount-chroot in the live-build working chroot dir. | 04:46 |
| infinity | Or livecd-rootfs/live-build/auto/build could trap exit and attempt a umount. | 04:48 |
| infinity | Maybe I'll give that a quick spin. | 04:50 |
| infinity | wgrant: Untested (yeehaw), but I think this should fix it: https://launchpad.net/ubuntu/+source/livecd-rootfs/2.302 | 05:14 |
| infinity | wgrant: Can you sort out exactly what it was trying to build, so we can map it to an API call to try to kill off some more nodes with testing? | 05:18 |
| wgrant | infinity: It was just a normal ubuntu-touch armhf build for wily. | 05:20 |
| infinity | https://launchpad.net/~ubuntu-cdimage/+livefs/ubuntu/wily/ubuntu-touch | 05:21 |
| infinity | Right, the lack of logs on those failures seems to match up with dead hosts. :P | 05:21 |
| === Odd_Blok1 is now known as Odd_Bloke | ||
| === doko_ is now known as doko | ||
| === sil2100_ is now known as sil2100 | ||
| sil2100 | Hello release team! I'll be disabling the image importer in nusakan for a short while if you don't mind | 16:05 |
| * ogra_ smells arale on the horizon | 16:17 | |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!