You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Currently, Nomad does not handle QEMU allocation migration properly. According to @diptanu
Yeah we simply shutdown the process. So no we don't shutdown the VM properly.
My use case specifically would be a Windows VMs and migrating those allocations between nodes. If I'm running a SQL Server database or IIS web server as an example, I wouldn't the VM process to just be killed and restarted on another node. I would almost prefer some form of "live migration" across hosts. The reason for this specifically is because most Windows software doesn't handle shutdowns very well or at all so just rapidly killing the process could have adverse effects. Another scenario would be when issuing a node-drain on a host.
The text was updated successfully, but these errors were encountered:
Hi @lfarnell! Just a heads-up: my team ran into this requirement as well, and we added a graceful shutdown option for the QEMU driver here (it uses the QEMU monitor under the hood): #3411
It's certainly not live migration, but it gives VMs the opportunity to shut down properly.
Currently, Nomad does not handle QEMU allocation migration properly. According to @diptanu
My use case specifically would be a Windows VMs and migrating those allocations between nodes. If I'm running a SQL Server database or IIS web server as an example, I wouldn't the VM process to just be killed and restarted on another node. I would almost prefer some form of "live migration" across hosts. The reason for this specifically is because most Windows software doesn't handle shutdowns very well or at all so just rapidly killing the process could have adverse effects. Another scenario would be when issuing a node-drain on a host.
The text was updated successfully, but these errors were encountered: