+++ This bug was initially created as a clone of Bug #1009179 +++ Description of problem: If the system changes its hostname while nova-compute is running, the resource update function will delete the compute_node record for the old hostname, rendering the host unusable by the scheduler until restart. Version-Release number of selected component (if applicable): How reproducible: Always Steps to Reproduce: 1. Start nova-compute 2. Change hostname/DNS such that that "virsh hostname" returns a different value 3. Note the compute record is deleted in the database after the next periodic update cycle Actual results: The compute_node record is deleted in the database and the scheduler fails to schedule instances on the node until nova-compute is restarted. Expected results: The node does not accept the changed name while running, delete the record, but rather warns the administrator about the detected change.
VERIFICATION: 1. Change hostname of the system to e.g. test.lab 2. Check out the /var/log/nova/compute.log 2013-12-02 11:15:18.897 8752 ERROR nova.virt.libvirt.driver [-] Hostname has changed from cougar14.scl.lab.tlv.redhat.com to test.lab. A restart is required to take effect. 3. The node does not accept the changed name while running and errors the administrator about the detected change.
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. http://rhn.redhat.com/errata/RHEA-2013-1859.html