If you use "args:" to add NICs, along with script=/path/to/script to attach them to the correct bridge, when you change networking on the pve host, these VMs become detached from the bridge, and the helper script is not re-executed, isolating the NICs.
The is only necessary for VMs that are distributed from vendors that specify driver+pci-but/slot combinations that are intened to run on rhel9 for example. Proxmox's version of qemu is newer, and does not assign pci-bus/slot combinations the same way that rhel's version does, making networking fail if you use the proxmox gui to attach vnic devices.
The ideal fix would be to have a setting in the GUI that would either emulate rhel9's pci-bus/slot assignments for NIC devices, or at least an option to hardcode pci-bus/slot in an 'advanced' NIC setting of some kind. Alternatively a way to re-trigger the helper script in the NIC definition within "args:" would also work.
The is only necessary for VMs that are distributed from vendors that specify driver+pci-but/slot combinations that are intened to run on rhel9 for example. Proxmox's version of qemu is newer, and does not assign pci-bus/slot combinations the same way that rhel's version does, making networking fail if you use the proxmox gui to attach vnic devices.
The ideal fix would be to have a setting in the GUI that would either emulate rhel9's pci-bus/slot assignments for NIC devices, or at least an option to hardcode pci-bus/slot in an 'advanced' NIC setting of some kind. Alternatively a way to re-trigger the helper script in the NIC definition within "args:" would also work.