2

I have installed successfully the Ubuntu landscape openstack autopilot installer. I created a couple of VM's successfully and can log into them by associating an IP.

Now here comes the issues, after I created a few vm's, every time I try now I get a waiting on network error. I have found fixes that state to add a timeout value to the nova.conf. Then when I search through all the LXC contains and find the nova.conf file, it say it is managed by juju and changes will be overwritten.

The real question I have is how do you troubleshoot an install like this? Its hard to know where everything is when it installs so many lxc containers and most I can't get into as there all ssh-key access. Is there any easier way to find things? I feel like I am making this harder than it should be. Any advice would be greatly appreciated.

Pabi
  • 7,401
  • 3
  • 40
  • 49
Chris Riley
  • 23

2 Answers2

3

I'd suggest looking over this question. It will direct you how to access the inner juju environment where you can start looking over things.

After you follow the above, you can explore your juju environment at a high level with juju status. I wouldn't suggest attempting to poke through machines and lxcs by hand without that high level tool. It would be very confusing indeed. Like navigating around a city without a map.

For instance, if you want to go to the nova-compute node, you type juju ssh nova-compute/0. Nice and easy, see juju help ssh for more details on ways to access the systems.

Also, many configuration options are tweakable through charm parameters:

juju get nova-compute
juju get nova-cloud-controller
juju set nova-cloud-controller "key=value"
juju ssh keystone/0

You get the idea.

Note also that Juju is not like Puppet. There is no periodic wipe of your settings files. It will only do that as a result of external actions. Like:

  • You running juju set
  • You rebooting the system (at reboot, it will probably reset values)
  • You relating another service to play with things/try things out.

I'd suggest reading more here:

Also the many links off those pages. Of course, those are all Ubuntu/Canonical/Juju links. I often find it helpful to follow up on the many general openstack resources when I need to dig in deeper. Which you seem to already be familiar with, so I'll spare you linking those here.

Good Luck!

dpb
  • 7,069
0

Are you talking about network errors on LXC containers that you created inside VMs that you spawned in the cloud? If yes, then it's an MTU problem and there is no automatic fix yet. You need to lower the MTU of the container interfaces to 1454. If you look at the VM interfaces, they already have that MTU.

Andreas Hasenack
  • 3,065
  • 1
  • 17
  • 17
  • THis is the error I get now when I try to create a new instance. Is the same error you are talking about? [link] https://sites.google.com/site/openstackinthebasement3/errors/net_error.JPG?attredirects=0[link] – Chris Riley Apr 19 '15 at 21:29
  • No, that's something else. You will have to check the nova-controller logs. – Andreas Hasenack Apr 20 '15 at 00:04