[09:39] <NuxRo> hi guys, I am trying to move configuration from scripts to cloud-init, in this case creating a swap file. Could anyone tell me what is wrong with my config file? https://paste.fedoraproject.org/467184/99307814/raw/
[09:40] <NuxRo> because no swap is produced
[09:40] <NuxRo> cloud-init 0.7.8-1-g3705bb5-0ubuntu1~16.04.3
[12:38] <smoser> NuxRo, that "works for me" here. in a quick test.
[12:38] <smoser> NuxRo, can you pastebin /var/log/cloud-init.log ?
[12:51] <NuxRo> smoser: sure, https://paste.fedoraproject.org/467252/47800466/raw/
[12:53] <smoser> hm.
[12:53] <smoser> that seems sparse.
[12:54] <smoser> cat you run:
[12:54] <smoser>  sudo cloud-init single --name=mounts --frequency=always
[12:54] <smoser> and paste that again ?
[12:57] <NuxRo> smoser: there you go https://paste.fedoraproject.org/467255/4929147/raw/
[12:57] <NuxRo> it complains about data source, but I do have it defined https://paste.fedoraproject.org/467257/14780050/raw/
[12:58] <smoser> NuxRo, well, it didnt find one... normally it'd find one and cache it to /var/lib/cloud/instance/
[12:59] <smoser> i have to run now. i can poke a bit later, where is this running ?
[12:59] <NuxRo> some cloudstack cloud in germany, can't disclose
[12:59] <NuxRo> thanks for the help
[13:26] <NuxRo> smoser: found the problem, pebkac as usual, template was generated on kvm, but booted on xen, NIC device naming got messed up. biosdevname=0 net.ifnames=0 to the rescue
[13:26] <NuxRo> once network correctly goes up, cloud-init functions as expected
[14:43] <smoser> NuxRo, hm..
[14:43] <smoser> cloud-init should be resilient to that.
[15:05] <NuxRo> well, clearly it's not, the VM had a working interface through which it could get to the datasource
[15:05] <NuxRo> this is ubuntu 16.04 with stock cloud-init
[15:13] <smoser> perhaps the instance id did not change while the network devices did
[15:58] <ThiagoCMC> Guys, how to tell cloud-init, on an Ubuntu 16.04 instance, to ignore the eth1/eth2/eth3/etc, and only configure eth0?
[15:59] <ThiagoCMC> At my Heat template, the eth0 network have dhcp, and eth1 doesn't but, cloud-init goes there and add a static IP for eth1!!! I want it to just ignore it but, how?
[16:02] <smoser> ThiagoCMC, this is openstack config drive i guess ?
[16:02] <ThiagoCMC> yes
[16:03] <smoser> You should be able to modify /etc/network/interfaces.d/50-cloud-init.cfg to your liking after the instance is up.
[16:04] <smoser> and it should not be broken
[16:04] <smoser> that is only written currently once per instance
[16:06] <ThiagoCMC> That's what I'm doing but, I would prefer to avoid that... So, there is no way to tell cloud-init to ignore that in first place?
[16:06] <smoser> ThiagoCMC, not right now there is not.
[16:07] <smoser> would you file a bug and describe your use case ?
[16:07] <ThiagoCMC> Sure!
[16:07] <ThiagoCMC> Here: https://launchpad.net/cloud-init ?
[16:10] <smoser> yeah
[16:10] <smoser> ThiagoCMC, this is someething i'm aware needs a policy, but i'm not sure on what the policy should be.
[16:11] <smoser> its kind of tricky...
[16:18] <ThiagoCMC> Ok, no problem, I'll fill a bug report, with details about this problem. Also publishing the Heat templates that I'm using, to simplify the reproduction.
[16:18] <ThiagoCMC> Thanks smoser!
[17:39] <NuxRo> If I wanted to override lock_passwd for default user, would a 99_unlock.cfg with this content suffice? https://paste.fedoraproject.org/467493/21945147/raw/
[17:45] <harlowja> smoser https://review.openstack.org/#/c/389324/ :-P
[17:45] <harlowja> u still in
[17:45] <harlowja> ha
[17:47] <smoser> \o/
[17:47] <smoser> i'm still important!
[17:49] <harlowja> :-P
[18:34] <NuxRo> well, that doesn't look like it did it, user still locked
[18:56] <smoser> NuxRo, http://paste.ubuntu.com/23412812/
[19:19] <NuxRo> cheers, was hoping I could get away with fewer lines than that :)
[19:20] <NuxRo> since default user is already defined in cloud.cfg, thought maybe just mentioning "default_user" would have been enough
[19:33] <robjo> smoser: Another topic, net/sysconfig/py is specific to RHEL, during implementation was there thought given about how this could/would be generalized or is this up to me to take it apart and put it back together in a different way?