[03:45] <thomasfedb> I have an Ubuntu VM on OpenStack with cloud-init. Although the instance is named "xyz.domain.com" in OpenStack, I find that /etc/hostname is just "xyz" rather than the FQDN. Is this expected behaviour? Configurable behaviour?
[13:21] <mbwe> i get a message in my cloud-init log, failed to shellify, that is with my rundcmd line
[18:44] <smoser> powersj: do you know about
[18:44] <smoser>  https://jenkins.ubuntu.com/server/job/cloud-init-ci/473/console
[18:44] <smoser> mbwe: can you give an  example of the runcmd ?
[18:45] <smoser> i've never seen that error and really wouldnt have expected that it could be buggy.
[18:45] <smoser> also paste your full /var/log/cloud-init.log would be hepful
[18:45] <smoser> helipful
[18:45] <smoser> helpful
[18:45] <smoser> third times a charm
[18:45] <powersj> smoser: Failed to destroy ZFS filesystem: cannot destroy 'default/containers/cloud-test-ubuntu-xenial-modules-set-password-list-t1tvm1dther3': dataset is busy
[18:45] <powersj> I have not seen that. I've seen something similar long ago
[18:46] <smoser> :-(
[18:58] <rharper> unlucky
[19:24] <smoser> rharper powersj :-(
[19:24] <smoser> twice in a row
[19:24] <smoser> https://jenkins.ubuntu.com/server/job/cloud-init-ci/474/console
[19:24]  * rharper asks smoser for lotto numbers
[19:24] <smoser> Failed to destroy ZFS filesystem: cannot destroy 'default/containers/cloud-test-ubuntu-xenial-modules-set-password-list-5rceznb9hsr9': dataset is busy\n"
[19:39] <rharper> smoser: https://github.com/zfsonlinux/zfs/issues/1810
[19:39] <rharper> interesting related to mounted namespaces in the containers
[19:41] <smoser> rharper: in #lxc-dev was talking to stgraber
[19:41] <smoser> there are 3 unmount processes eathing 100% of cpu on that system now
[19:41] <smoser> (representing 3 contaienrs that failed to die)
[19:42] <rharper> eww
[19:42] <rharper> I wonder if there is a new systemd in xenial recently
[20:12] <msaikia> Hi, can anyone please take a look at this changeset
[20:12] <msaikia> https://code.launchpad.net/~msaikia/cloud-init/+git/cloud-init/+merge/330105
[20:35] <smoser> rharper: how do i run lp:~raharper/curtin/trunk.vmtests-v3-streams on diglett
[20:36] <smoser> youo have imates there ?
[20:36] <rharper> smoser: one sec
[20:36] <rharper> IMAGE_DIR=/srv/tmp/rharper/images
[21:35] <blackboxsw> smoser: I wonder why lxc doesn't set hostname fqdn in /etc/hosts as our integration tests seem to expect.
[21:35] <blackboxsw> tried hostname --fqdn on the lxc instance when providing cloud-config   hostname: def\n fqdn: abc.domain.com and hostname --fqdn is alsways lxd.
[21:43] <blackboxsw> your cii-fixup branch is approved. msaikia I'm on your branch now for real
[21:48] <smoser> blackboxsw: i dont think i follwo.
[21:48] <blackboxsw> smoser: np I know you're working something else. Getting a pastebin together.
[21:52] <blackboxsw> http://pastebin.ubuntu.com/25906048/
[21:53] <blackboxsw> smoser: ^ I expected to see the fqdn set (or at least hostname --fqdn as we test in integration tests)
[21:55] <blackboxsw> I was trying to validate what our integration test is also 'testing'   in tests/cloud_tests/testcases/modules/set_hostname_fqdn.py
[21:56] <blackboxsw> I also note that redhat behaves differently than ubuntu in this regard. if hostname and fqdn are provided. redhat/cent uses fqdn, ubuntu _select_hostname() returns just the hostname when both are provided
[21:56] <blackboxsw> on first blush.
[21:57] <blackboxsw> maybe I'm wrong there.
[22:36] <smoser> blackboxsw: i'm going to pull ci
[22:36] <smoser> and yes, hostname is messed up
[22:37] <smoser> due largely to silly things like sysatemd-hostnamectl and sudo
[22:41] <powersj> smoser: can you let me test this evening/tomorrow morning?