Ceph HA Cluster automatic failover downtime

[iPero]

Hi guys, i was setting up an HA cluster for automatic failover and succeded, the only issue here is the downtime in case one of the nodes fails.
I need to have a service up and running with near to 0 downtime even if a machine crash for whatever reason, right now it takes about 2 minutes for the cluster to recognized the node as failed and migrate the vm to another node, is there any way to reduce this timout?

Thanks,
Tommaso.

 

[ph0x]

Not on the hypervisor level. Aim for service level redundancy if you need less downtime, e.g. a failover machine.