Can't deploy service in a freshly-bootstrapped LXC environment

Bug #1073308 reported by David Owen
6
This bug affects 1 person
Affects Status Importance Assigned to Milestone
pyjuju
Triaged
Low
Unassigned

Bug Description

On a freshly-bootstrapped LXC environment, trying to deploy a new service will allocate an instance which stays in the "pending" state indefinitely.

1. I'm using juju 0.6.0.1 for Ubuntu 12.04.
2. Make sure that a) the directory containing the local environment and b) /var/lib/lxc are both completely empty.
3. Reboot.
4. "juju boostrap"
5. "juju status" will display the new environment
6. "juju deploy <charm>"
7. "juju status" will show output similar to status.txt, attached.

I'm not sure where to look for potential problems, but I also found some interesting content in the unit's container.log (attached). output.log is linked to a file containing only "/usr/bin/python: No module named juju.agents" several times. unit.log is a link to a non-existent file.

Revision history for this message
David Owen (dsowen) wrote :
Revision history for this message
David Owen (dsowen) wrote :
description: updated
Revision history for this message
David Owen (dsowen) wrote :

I just found some failure messages in /var/log/cloud-init{,-output}.log inside the LXC container. I don't know if anything here is pertinent.

Revision history for this message
David Owen (dsowen) wrote :
Revision history for this message
Seth Arnold (seth-arnold) wrote :

David, I found my first attempts at juju deployment failed with the same error message you have; thanks for giving the /var/log/cloud-init-output.log name within the container, that was sufficient for me to determine that my apt-cacher-ng was not yet configured to listen on the rfc1918 IP address that LXC or Juju configured for me.

Your log file shows similar failures to retrieve packages; it looks to me like your local package mirror is broken or out of date or similarly busted.

Hope this helps.

Revision history for this message
David Owen (dsowen) wrote :

I don't think there's a problem with the address that apt-cacher-ng is listening on. After purging apt-cacher-ng and reinstalling, I see that the LXC container is able to fetch index files through the cache, but then there are size mismatches on basically every deb file.

Revision history for this message
David Owen (dsowen) wrote :

Besides what may or may not be wrong with apt-cacher-ng, juju should not lose track of an instance in this way; it should detect that the instance has failed in someway and report that to the user.

Revision history for this message
David Owen (dsowen) wrote :

Tried clearing the cache *and* disabling ufw, as described in http://askubuntu.com/questions/134977/juju-stuck-in-pending-state-when-using-lxc . No luck.

Curtis Hovey (sinzui)
Changed in juju:
importance: Undecided → Low
status: New → Triaged
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.