We are trying to build HA cluster with iSCSI storage over dedicated network interface.
Looks like disconnecting storage network (in any way - physical fault or misconfiguration) doent lead to any reaction from PVE HA.
For example, if node connection to iscsi storage failed, we got messages like
But VMs (protected by HA) still in "Running" state, regardless of their real state (remounted in readonly FS or i/o errors).
So i lis there a way to force migrate VMs with faulty storage.
Or, at least, a way to mark node as faulty and do "fenccing" with Proxmox embedded ways when icmp ping to storage fails.
Looks like disconnecting storage network (in any way - physical fault or misconfiguration) doent lead to any reaction from PVE HA.
For example, if node connection to iscsi storage failed, we got messages like
Code:
May 18 20:06:28 pvx iscsid[1356]: connect to 172.16.100.100:3260 failed (No route to host)
May 18 20:06:33 pvx pvestatd[1796]: storage 'iscsi-volume' is not onlineSo i lis there a way to force migrate VMs with faulty storage.
Or, at least, a way to mark node as faulty and do "fenccing" with Proxmox embedded ways when icmp ping to storage fails.