Hi,
We are setting up a brand new PVE instance dedicated to our firewalls stacks.
All those FWs rely on VFs for there network interfaces.
The old solution was based on libvirt, and VFs could be dynamically attributed to VMs.
To mimick this behaviour I've setup an pre-start hook script that get info from the config.description and then setup the VF accordingly .
It check that the VF is not assigned to any VMs then feed the config file.
However, it seems that even if this a pre-start, configuration is already loaded from file and modification are not applied to the VM to be run.
I've also tried to use "qm config set" or "pvesh set" without any success as then both try to lock the VM and so, fail.
Is there a better way to handle that ?
Why does with pre-start the configuration is already lock ?
We are setting up a brand new PVE instance dedicated to our firewalls stacks.
All those FWs rely on VFs for there network interfaces.
The old solution was based on libvirt, and VFs could be dynamically attributed to VMs.
To mimick this behaviour I've setup an pre-start hook script that get info from the config.description and then setup the VF accordingly .
It check that the VF is not assigned to any VMs then feed the config file.
However, it seems that even if this a pre-start, configuration is already loaded from file and modification are not applied to the VM to be run.
I've also tried to use "qm config set" or "pvesh set" without any success as then both try to lock the VM and so, fail.
Is there a better way to handle that ?
Why does with pre-start the configuration is already lock ?
Attachments
Last edited: