prevent vm to start on same node (HA)

[ugf]

Hello,

I have a setup with shared storage set up and 5 nodes.

If one node crashes the VMs should be instant migrated on another node and crashed node should be left empty.

My problem :

1. It takes really long for proxmox to actually start the HA-Migration (e.g. If i plug out power cable while VMs are running)
2. If the node just reboots (e.g. after oopsie), the vms will be started again on that node as they will wait so long (see 1.)

How can I let the VMs get migrated faster and also make sure that the host does not start the vms again?

 

[sb-jw]

If one node crashes the VMs should be instant migrated on another node and crashed node should be left empty.

High availability usually means an availability of at least 99.9% - but even then that means up to 8:45:58 hours of downtime per year. Based on this, Proxmox’s response time is reasonable.

Another distinction to be made is that Proxmox can only do HA and has no fault tolerance.

When you put a VM in HA mode, you can set the behavior in the groups. For example, there is the nofailback option there.

See also docs: https://pve.proxmox.com/pve-docs/pve-admin-guide.html#_configuration_16

 

[bbgeek17]

How can I let the VMs get migrated faster and also make sure that the host does not start the vms again?

If things did not change significantly, then you are looking at 2min wait time for cluster to make a decision on host being down and moving the VM
https://forum.proxmox.com/threads/faster-failover-possible.36894/#post-180841

https://pve.proxmox.com/wiki/High_A...you can get no more than 99.999% availability.

If your host boots significantly faster than 2 minutes, then the VM will not be migrated as currently designed.
With planned maintenance this means that you should manually migrate VMs over before taking the host down.

---

Blockbridge : Ultra low latency all-NVME shared storage for Proxmox - https://www.blockbridge.com/proxmox