| 2018-10-18 09:36:35 |
Lucian Petrut |
description |
In some cases, after instances are failed over, the destination host cannot find the instance vNICs or switch ports.
It looks like the ports are not ready yet by the time we get the failover event. Adding a retry to the method that connects ports to vswitches seems to avoid this issue.
Trace: http://paste.openstack.org/raw/730451/ |
In some cases, after instances are failed over, the destination host cannot find the instance vNICs or switch ports.
It looks like the ports are not ready yet by the time we get the failover event. Adding a retry to the method that connects ports to vswitches seems to avoid this issue.
Trace: http://paste.openstack.org/raw/730451/
Later edit: This is caused by the fact that we're not waiting for "pending" cluster groups, in which case the VMs aren't registered in Hyper-V yet. |
|
