How to live migrate VMs?
- Thread starter Mayank006
- Start date
One option is to right click on the running VM and select "Migrate" option.
Blockbridge : Ultra low latency all-NVME shared storage for Proxmox - https://www.blockbridge.com/proxmox
Blockbridge : Ultra low latency all-NVME shared storage for Proxmox - https://www.blockbridge.com/proxmox
I was powering off one of the servers and was expecting that VMs should live migrate automatically. Does proxmox support this automatically?
My Vms are restarting, which results in 30-40 sec of downtime.
Sounds like you were not powering off, but rebooting?
I recommend going over this documentation for some explanations and proper expectations:
https://pve.proxmox.com/pve-docs/ch... moved back.-,Shutdown Policy,-Below you will
Blockbridge : Ultra low latency all-NVME shared storage for Proxmox - https://www.blockbridge.com/proxmox
I recommend going over this documentation for some explanations and proper expectations:
https://pve.proxmox.com/pve-docs/ch... moved back.-,Shutdown Policy,-Below you will
Blockbridge : Ultra low latency all-NVME shared storage for Proxmox - https://www.blockbridge.com/proxmox
I am doing tests like removing the power plug from one out of 3 servers to see how HA works. I noticed that when I do manual migration it is live migration, but when Vms automatically starts migration after removing the power plug from the server then is not live migration.
This is called "Fault-tolerance", and it's not supported by proxmox currently.
A qemu project exist for this:
https://wiki.qemu.org/Features/COLO
but AFAIK, it's incomplete (don't support shared storage)
Live: a manual migration is an actively managed and controlled operation. "Everything" is moved to the destination, including the active RAM and the CPU state. The main process on the source is stopped and restarted on the destination. It got not stopped, at least not in a logical sense.
High available: When you unplug the power that same process dies immediately. The content of the RAM is lost. The current CPU state of the the VM is lost. The complete VM is turned off, brutally. HA notices this after a minute or so and starts a fresh instance of that VM on an alternate node...
Best regards