I have CEPH and HA configured and working properly, however in testing when I shutdown a physical node the VM is migrated and rebooted what I want to understand is if this is normal behavior or if there is a way not to reboot the VM when the physical the node fails
CEPH and HA switching without restarting VM
- Thread starter Servnet Mexico
- Start date
-
- Tags
- ceph high avaibility
Thank You,
Is there any URL where this operation is described?
for example, in VmWare when the node dies, the VM migrates and does not restart.
Is there any URL where this operation is described?
for example, in VmWare when the node dies, the VM migrates and does not restart.
If the VM is not under hot-standby (where memory is actively synced), then even VMWare needs to start (recover) the VM on a different node.
I agree, the node isn't really "dieing" if its able to be live migrate VM's to another node. Most real failures cause a node to go down hard.
This would be a good read for you.
https://pve.proxmox.com/wiki/High_Availability
This would be a good read for you.
https://pve.proxmox.com/wiki/High_Availability
Sorry to resurrect this old thread but I thought it might benefit others that come across it.
I was not aware that always-on or hot-standby was not a feature of Proxmox HA and only discovered this after commissioning 8 servers in a cluster with CEPH. I made the mistake of assuming that this function would be possible since we've had this working on our current environment for quite a few years.
Perhaps the docs could more clearly explain that parallel VM's/hot-standby is not possible with Proxmox? Although entirely my fault, having discovered this limitation this late in the process has come as somewhat of a shock and I'm still trying to figure out what my next course of action will be.
Our current solution, although proprietary and more expensive than Proxmox, has worked perfectly in terms of seamless failover of VM's and is Linux based using KVM. For us the main reason for wanting to switch from our current solution was the cost and also the greater control that Proxmox gives in general in terms of custom configurations.
I was not aware that always-on or hot-standby was not a feature of Proxmox HA and only discovered this after commissioning 8 servers in a cluster with CEPH. I made the mistake of assuming that this function would be possible since we've had this working on our current environment for quite a few years.
Perhaps the docs could more clearly explain that parallel VM's/hot-standby is not possible with Proxmox? Although entirely my fault, having discovered this limitation this late in the process has come as somewhat of a shock and I'm still trying to figure out what my next course of action will be.
Our current solution, although proprietary and more expensive than Proxmox, has worked perfectly in terms of seamless failover of VM's and is Linux based using KVM. For us the main reason for wanting to switch from our current solution was the cost and also the greater control that Proxmox gives in general in terms of custom configurations.