instance/volume creation error

asked 2013-12-03 23:14:59 -0500

dtaylor gravatar image

Hello all,

I am hoping some of you can help point me in the right direction. I am building a proof of concept Openstack setup using GlusterFS as back-end storage for virtual machine volumes. I am relatively new to Openstack, so I am not sure what my problem is.


1 x Controller node (Keystone, Glance, Nova Controller, Cinder, Heat, Ceilometer, Neutron Server) 1 x Compute node (Nova Compute, Cinder, Neutron Agent) 1 x Network node (Neutron dedicated server) 3 x GlusterFS nodes (2 exported volumes each, one for Glance images/Nova instances, one for Cinder volumes)

Volume 0: Mounted at (/var/lib/cinder/volumes) Volume 1: Mount at (/csv1) with nova/glance subdirectories owned by the appropriate user/groups

I am using CentOS 6.5 on all nodes running the 2.6.32-431 kernel. Currently the firewall is disabled on all nodes and SELinux is set to enforcing. I have two networks, an internal and external network. All nodes are resolvable with forward/reverse DNS.

Internal Network: External Network:

Communication using Qpid between nodes works with no errors, MySQL connections are good.

Here is my problem:

When creating a new instance in Horizon, booting from an existing image and creating a new volume, instance creation always fails at the block device setup stage. If I boot directly off the image without creating a new volume, the instance starts an I am able to connect via the VNC console.

I am booting from a CentOS 6.5 minimal ISO using Glance, stored on a separate GlusterFS volume that Glance is configured to use as the 'state_path'. I can create images fine on this share.

After instance creation fails, if I try to access the 'Volumes' tab in Horizon, I am always given an internal error message:

AttributeError at /project/volumes/
Request Method: GET
Request URL:
Django Version: 1.4.8
Exception Type: AttributeError
Exception Value:    
Exception Location: /usr/lib/python2.6/site-packages/cinderclient/ in __getattr__, line 271
Python Executable:  /usr/bin/python
Python Version: 2.6.6
Python Path:    
Server time:    Wed, 4 Dec 2013 04:43:27 +0000

In order to acces the Volumes page again I have to manually delete the Cinder volume from the command line using 'cinder delete'. Once the volume is gone I can access the page again. The volumes path is located on a GlusterFS share. This is my 'cinder.conf' file:


# SQL Connection

# Metering Configuration
notification_driver=cinder ...
edit retag flag offensive close merge delete

1 answer

Sort by ยป oldest newest most voted

answered 2013-12-06 01:59:36 -0500

dtaylor gravatar image

I think this is related to an existing bug - .

edit flag offensive delete link more

Get to know Ask OpenStack

Resources for moderators

Question Tools

1 follower


Asked: 2013-12-03 23:14:59 -0500

Seen: 1,347 times

Last updated: Dec 06 '13