[13:28] <smoser> msaikia, so rebase to master
[13:28] <smoser> that make sense ?
=== rangerpbzzzz is now known as rangerpb
[18:38] <rangerpb> smoser, paulmey i cobbled up a patch where we call perform_hostname_bounce when using only cloud-init to provision; however, i observe later that both the cc_set_hostname and update_hostname override that with incorrect results.
[18:38] <rangerpb> should those be called in an azure deployment ?
[18:39] <rangerpb> i think part of the problem is they aren't able to detect a proper hostname or fqdn for some reason ... been poking around at the code... seems the best way to set the hostname is to pluck it from the cdrom's metadata
[18:40] <smoser> something is wrong if they're setting it wrong.
[18:40] <smoser> they should get the write hostname from the metadata and set it.
[18:40] <rangerpb> so both of those should run in your opinion smoser?
[18:41] <smoser> i think so, yeah.
[18:41] <rangerpb> of course, part of the issue is the the hostname stuff boils down to a method in util.py which is unaware of the metadata afaict
[18:42] <rangerpb> is there any difference in setting hostname with hostname vs hostnamectl ?
[18:44] <smoser> well, i think there is a bug with setting it with hostnamectl
[18:45] <smoser> possibly that is what you're seeing.
[18:45] <smoser> cloud-init possibly tries to set it to early ? larsks mentioned something about this.
[18:45] <larsks> I'm not sure about too early.  There was an issue concerning cloud-init, hostnamectl, and dbus...but that results in "it all asplode" rather than failure to set the hostname.
[18:46] <larsks> But rangerpb I agree with smosers: the distro update_hostname method should do the right thing.
[18:46] <larsks> That means that the data source needs to provide the correct host name.  If it's not, that's where the problem is happening.
[18:46] <rangerpb> distro update_hostname ?
[18:47] <rangerpb> what do you mean provide it ?
[18:47] <larsks> rangerpb: unless I am crazy, the update_hostname method is provided by the distro class(es).
[18:47] <rangerpb> maybe i am missing something
[18:47] <larsks> And the data source driver should be providing those methods with the correct hostname.
[18:48] <larsks> I have to run to a parent/teach conference.  Back in a bit!
[18:49] <rangerpb> well there is update_hostname defined in distros/__init__.py and there is a cc_update_hostname.py
=== rangerpb is now known as rangerpbzzzz