Hello. We are trying to set up KVM HA (Host-HA & VM-HA). When host goes down (for example: forcibly reset, unclean shutdown, stalling the next boot), all the normal fencing procedures go on as one would expect. Host gets put in maintenance mode and everything.
However, the instances remain unmigrated until the affected machine is back online - even if the machine faces difficulties getting back online. Can this behavior be changed or is there a possible configuration error on our side? We would expect VMs to be a first priority to get back online and host to be quarantined on such event. Currently it looks like HA cares more about getting host back online than customer workloads. OOBM access is set up via Redfish. Primary storage is Shared Mount Point (GFS2), HA heartbeat via NFS. Any materials/advice on this matter is most welcome. Lugupidamisega / Best regards, Joann Mõndresku Süsteemiadministraator | Systems Administrator | jo...@wavecom.ee WaveCom AS | ISO 9001, 27001 & 27017 Certified DC and Cloud services Endla 16, Tallinn 10142 | www.wavecom.ee | www.facebook.com/wavecom.ee