List of Common Errors (and Solutions) while Installing and Configuring OpenStack Nova Service

In continuation to my previous articles on OpenStack Errors and Solutions for installation and configuration of KeyStone and Glance, today I’m going to share some errors that I encountered while installing and configuring Nova (Compute) service on controller node.

Most of these errors were encountered while listing the images via nova command, starting various nova services such as (nova-api, nova-metadata-api, nova-cert, nova-conductor, nova-console, nova-consoleauth, nova-scheduler etc…). The errors occurred while starting the nova services were bit tricky, it was mainly due to improper configurations and syntax errors.

Ok, Here’s the list of problems that I faced.

Solutions for openstack errors

ERROR: (‘Connection aborted.’, error(111, ‘Connection refused’))

# nova image-list
ERROR: ('Connection aborted.', error(111, 'Connection refused'))


When I tried to image-list via nova, the connection refused error was thrown. This was primarily due to various reasons,

  • Check if all the nova related services are running properly. For example, when you start any nova services as ‘service nova-api start‘, the command might return green status as “[OK]’. But actually the service might not have been started properly. Below is one example.
# /etc/init.d/openstack-nova-api start
Starting openstack-nova-api:                               [  OK  ]
# /etc/init.d/openstack-nova-api status
openstack-nova-api dead but pid file exists

In above case, you should check the corresponding service log under /var/log/nova/api.log to know the exact error message.

In my case, none of the nova service were starting properly and I had to dig all of those service logs.

openstack-nova-api is running...
openstack-nova-cert is running...
openstack-nova-conductor is running...
openstack-nova-console is running...
openstack-nova-consoleauth is running...
openstack-nova-metadata-api is running
openstack-nova-novncproxy is running...
openstack-nova-scheduler is running...
  • Check if the firewall on the controller node is blocking the service ports.

Error : CRITICAL nova [-] RequiredOptError: value required for option: lock_path

The above error message was captured from /var/log/nova/api.log file and it was triggered when openstack-nova-api service failed to start.


# vim /etc/nova/nova.conf

and uncomment ‘lock_path‘ : lock_path=/var/lib/nova/tmp. Save the file and start the service.

ERROR nova.openstack.common.threadgroup [-] [Errno 13] Permission denied: ‘/usr/lib/python2.6/site-packages/CA’

The above error message was logged in /var/log/nova/cert.log and it was triggered when ‘openstack-nova-cert‘ failed to start.


# vim /etc/nova/nova.conf

and check if ‘ca_path‘ is pointing to the right directory and user ‘nova’ has permission to it.

Also Read: How to Fix OpenStack’s Keystone Authentication Error – HTTP 500?

ERROR nova.wsgi [-] Could not bind to
CRITICAL nova [-] error: [Errno 98] Address already in use

# service openstack-nova-metadata-api status
openstack-nova-metadata-api dead but pid file exists

And /var/log/nova/metadata-api.log reported the above error message.


Surprisingly, the service ‘openstack-nova-metadata-api‘ was already running. Because, ‘openstack-nova-api‘ starts ‘metadata-api‘ service along with it and thus the service ‘metadata-api‘ fails to bind to the port. All you need to do is, check if ‘nova-api’ is configured to start ‘metadata-api’ along with it. To do that,

# vim /etc/nova/nova.conf and lookout for ‘enabled_apis = osapi_compute,metadata

If you find metadata in enabled_apis, then whenever nova-api is started, it will also start metadata api.

In case, if you want to start nova-metadata-api individually, then remove metadata from ‘enabled_apis’.

'enabled_apis = osapi_compute'
# service openstack-nova-api start
# service openstack-nova-metadata-api start

Now the service should start individually.

Error: compute driver option required but not specified

# service openstack-nova-compute status
openstack-nova-compute dead but pid file exists


#vim /etc/nova/nova.conf and set ‘compute_driver‘.

TRACE nova.openstack.common.threadgroup OSError: [Errno 2] No such file or directory: ‘/usr/lib/python2.6/site-packages/instances’

# service openstack-nova-compute status
openstack-nova-compute dead but pid file exists

#vim /etc/nova/nova.conf and set ‘instances_path=/var/lib/nova/instances

You might also be interested in reading : List of Most Common Errors (and Solutions) While Installing OpenStack Image Service GLANCE

And the Bonus is here…

Related Content
vps germany
Disclaimer: The content published in this article is the views of the author only. Techglimpse does not gurantee accuracy, completness or validity. If you believe the content on this post violates your copyright, please send us a mail for removal. Read more.