Instances do not run because /var/run/eucalyptus/net is missing

[Thierry Carrez]

Running 20090929.2 UEC cloud (1.6~bzr854-0ubuntu11), MANAGED-NOVLAN with public IPs defined
Starting an instance fails. It goes from pending to running to shutting-down to terminated.

In cc.log:
failed to (re)create DHCP config
(//var/run/eucalyptus/net/euca-dhcp.conf)
[Tue Sep 29 16:37:08 2009][007346][EUCAERROR ] cannot start DHCP daemon, for instance i-4A9A0861 please check your network settings

Creating /var/run/eucalyptus/net manually makes it work.

[Thierry Carrez (ttx)]

Init script was running /usr/sbin/euca_conf --check cc before starting the cc.
This would create the missing directory and do lots of other small things.
This is missing from the upstart script.

[Thierry Carrez (ttx)]

Assigning to the US shift :)
This needs to be in a beta respin asap.

[Dustin Kirkland  (kirkland)]

I've committed a fix, r637, r638. Tested these, and the dirs are now being created.

Upload pending a couple of other bug fixes to be batched.

:-Dustin

[Launchpad Janitor (janitor)]

This bug was fixed in the package eucalyptus - 1.6~bzr854-0ubuntu12

---------------
eucalyptus (1.6~bzr854-0ubuntu12) karmic; urgency=low

  * eucalyptus-cc.upstart, eucalyptus-common.eucalyptus.upstart: ensure
    that the /var/run/eucalyptus/net directory gets created on upstart
    job runs, and that all of /var/run/eucalyptus is owned by
    eucalyptus:eucalyptus, LP: #438747
  * debian/eucalyptus-common.postrm: fix lintian error, add missing
    debhelper stubb

 -- Dustin Kirkland <email address hidden> Tue, 29 Sep 2009 15:55:45 -0500

[Thierry Carrez (ttx)]

Don't change status without a comment explaining why.