OpenStack doesn't apply default configurations anymore

asked 2018-12-15 05:17:06 -0600

soheil08 gravatar image

updated 2018-12-15 05:21:57 -0600

Hi, I have configured an OpenStack 2 nodes cluster. One node acts as the controller and compute node and the other as just compute node. The cluster was working fine until I activated a new NIC interface and add it to the property physical_interface_mappings: physical_interface_mappings = provider:enp2s0,provider1:enp7s4 Since then some services only on the controller node like glance and controller nova services stop to working properly.

It seems the services cannot detect default configuration anymore. For example, the glance error was about Unable to locate paste config file for glance-api and when I set the config_file=/usr/share/glance/glance-api-dist-paste.ini in the file glance-api.conf, it disappeared. This is while before activating the new NIC interface, it worked without any needs to config_file property! On the other hand, the nova-compute service errors about cannot find instances directory and set the property state_path=/var/lib/nova. This is while before activating new NIC interface it used the path /var/lib/nova as state path and it had created directories like instances there! The compute service on the controller node didn't detect the driver and error. I set the property compute_driver=libvirt.LibvirtDriver and it solved.

In addition, most of the services don't write logs on the log file anymore! For example, nova-compute didn't write any log into its log file since I rebooted the controller node! this is while the compute service is up and running on the controller node!

What can be the cause of all new errors and behavior?

edit retag flag offensive close merge delete

Comments

Have you tried removing the extra physical interface mapping to verify that it is responsible for the issues? What are you trying to do that has you adding a secondary physical provider interface?

reynoni gravatar imagereynoni ( 2018-12-20 11:25:00 -0600 )edit