Ask Your Question

Revision history [back]

click to hide/show revision 1
initial version

It depends. For e.g a) Nova Compute: then you can configure the triggering of evacauation and migration procedure of the VMs to an healthy compute node. You can read more about it here: https://docs.openstack.org/nova/rocky/admin/evacuate.html

b) Controllers: They run in a quorum and by deafult there are 3 running instances. So failure of any one of the server causes no issues.

c) Storage ( for e.g. volume storage in CEPH): Here the data is duplicated to additional server nodes in a cluster by replication or erasure coding. In ceph by default there are three server nodes in a cluster. So even if any one of the nodes fails, VM can still find its data from the remaining two health nodes of the cluster