Ask Your Question
0

Glance image delivery not working on compute nodes

asked 2011-07-09 14:34:21 -0500

howard-h gravatar image

Hi all,

Glance image delivery doesn't seem to be working for me!

I have a three server set-up with Glance running on the cloud controller which also runs compute. Instances that start on the cloud controller work fine but those that attempt to start on the compute nodes always fail. I have the same nova.conf on the nodes and controller. The relevant part of the compute log on the node is:

2011-07-09 15:07:52,811 DEBUG nova.utils [-] Running cmd (subprocess): sudo tune2fs -c 0 -i 0 /dev/nbd15 from (pid=1324) execute /usr/lib/pymodules/python2.7/nova/utils.py:150 2011-07-09 15:07:52,856 DEBUG nova.utils [-] Result was 1 from (pid=1324) execute /usr/lib/pymodules/python2.7/nova/utils.py:166 2011-07-09 15:07:52,857 DEBUG nova.utils [-] Running cmd (subprocess): sudo qemu-nbd -d /dev/nbd15 from (pid=1324) execute /usr/lib/pymodules/python2.7/nova/utils.py:150 2011-07-09 15:07:52,906 WARNING nova.virt.libvirt_conn [-] instance instance-00000018: ignoring error injecting data into image 2 (Unexpected error while running command. Command: sudo tune2fs -c 0 -i 0 /dev/nbd15 Exit code: 1 Stdout: 'tune2fs 1.41.14 (22-Dec-2010)\n' Stderr: "tune2fs: Invalid argument while trying to open /dev/nbd15\nCouldn't find valid filesystem superblock.\n") 2011-07-09 15:08:25,952 ERROR nova.exception [-] Uncaught exception (nova.exception): TRACE: Traceback (most recent call last):

Additionally, the files in /var/lib/nova/instances/_base are zero length; I've seen that others have copied the contents of the controller's version of this directory into here but this seems to defeat the object of Glance.

Also, "glance index --host=10.0.0.100" on a node gives the following error (it works fine on the controller):

Failed to show index. Got error: Unknown error occurred! {"versions": [{"status": "CURRENT", "id": "v1.0", "links": [{"href": "http://0.0.0.0:9292/v1/", "rel": "self"}]}]} howard@stratus:~$ ping 192.168.1.44

Can anyone give me some suggestions of a solution or further diagnostic steps--others have posted the same problem but I've not seen a solution.

Thanks

edit retag flag offensive close merge delete

23 answers

Sort by ยป oldest newest most voted
0

answered 2011-07-29 20:25:18 -0500

jaypipes gravatar image

Hi Howard,

I'm not sure what I can do here. There's been some sort of corruption of the registry database, where the db_sync command is bombing out. The corruption is due to the registry database apparently not having the correct database version record in the migration table. I'm not sure what I can recommend other than re-installing Glance's database. Perhaps the only other solution would to make a tarball of the SQL dump of your glance registry, email me the tarball and I can try to figure out how to uncorrupt it.

-jay

edit flag offensive delete link more
0

answered 2011-08-11 03:30:01 -0500

we faced the same issue and after following Jay's advice, we issued the following commands to Compute node and then it worked! (so that now we no longer need to manually move /var/lib/nova/instances/_base from Controller node to Compute node to launch/run instances on Compute node.)

sudo apt-get purge nova-compute sudo apt-get install glance sudo purge nova-common sudo apt-get install nova-compute mkdir /var/lib/nova/instances chmod 777 /var/lib/nova/instances sudo /etc/init.d/libvirt-bin restart sudo restart nova-compute

(note: we didn't do sync_db on Compute node)

daniel

edit flag offensive delete link more
0

answered 2011-08-11 18:00:33 -0500

forgot to mention that we're using "diablo d3" from the ppa:nova-core/milestone

edit flag offensive delete link more

Your Answer

Please start posting anonymously - your entry will be published after you log in or create a new account.

Add Answer

Get to know Ask OpenStack

Resources for moderators

Question Tools

1 follower

Stats

Asked: 2011-07-09 14:34:21 -0500

Seen: 426 times

Last updated: Aug 11 '11