How to reduce the timeout when migrating a VM in a Proxmox cluster when a node fails?

Question

Debian

Good afternoon
I'm trying to prepare a cluster of 3 nodes based on GlusterFS and Proxmox cluster for moving some services, HA is enabled on the proxy for the test virtual machine, migration is in progress. But the problem is different, if the connection with the node is lost for a long time, you have to wait until the cluster perceives the node as definitely non-working and starts the virtual machine on the next one. It takes 2 minutes before the start of the virtual machine starts, plus 30 seconds to start it, for a total of 2 minutes 30 seconds of downtime, which is quite a lot. Can this timeout be somehow reduced?

Answer 1

I dug around on the proxmox forum, at the moment, apparently, these timeouts are necessary for their fence algorithm to work, there was talk about the appearance of some options in future versions that will help reduce this time. Well, in general, very much so, but oh well.

Answer 2

but how do you reduce it, in fact, it takes time for fencing, and if you do everything instantly, then you can grab two equally running virtual machines