I added an SDN zone for vlan -> vmbr0 (vlan aware), then added vnets with id of 100, 1000 (matching the vlans in my physical system) and things work with vm's getting DHCP addresses, etc.
However, adding a vlan with id=1 causes the network to fail and loss of management access/etc.
only the first node getting the SDN update is lost as the other nodes show SDN "pending" since (I believe) there is never a conclusion to the SDN apply for that node.
I have side access through an frr ip6 link from an unimpacted node and can ssh to the lost node.
ip a shows the link is up, but it can only ping itself successfully.
e.g., (it's ip4 addr is 192.168.0.81)
ping 192.168.0.81 works, but pinging the gateway, or its sibling nodes .1, .82, .83 all fail.
The only way back is the edit /etc/network/interfaces and comment out the source /etc/network/interfaces.d/* line and reboot or restart network services.
Then clean up things to remove the vlan vnet with id=1.
The goal was to create an SDN named virtual switch that simply connected to vmbr0 (analogous to a port group in ESXI). This would allow vlan control of a group of VMs without having to manually go into each and set a vlan id.
I know I can just connect a vm to vmbr0 - that is not the point.
The issue is , ID = 1 apparently is a valid input in the interface, and it borks the network interface.
I will note that by default vmbr0 being vlan aware does have: bridge-vids 2-4094
Changing the 2 to a 1 does not appear to be a valid option.
However, adding a vlan with id=1 causes the network to fail and loss of management access/etc.
only the first node getting the SDN update is lost as the other nodes show SDN "pending" since (I believe) there is never a conclusion to the SDN apply for that node.
I have side access through an frr ip6 link from an unimpacted node and can ssh to the lost node.
ip a shows the link is up, but it can only ping itself successfully.
e.g., (it's ip4 addr is 192.168.0.81)
ping 192.168.0.81 works, but pinging the gateway, or its sibling nodes .1, .82, .83 all fail.
The only way back is the edit /etc/network/interfaces and comment out the source /etc/network/interfaces.d/* line and reboot or restart network services.
Then clean up things to remove the vlan vnet with id=1.
The goal was to create an SDN named virtual switch that simply connected to vmbr0 (analogous to a port group in ESXI). This would allow vlan control of a group of VMs without having to manually go into each and set a vlan id.
I know I can just connect a vm to vmbr0 - that is not the point.
The issue is , ID = 1 apparently is a valid input in the interface, and it borks the network interface.
I will note that by default vmbr0 being vlan aware does have: bridge-vids 2-4094
Changing the 2 to a 1 does not appear to be a valid option.