Additional networks need routing so that they can also use the sw1 tunnel rather than adding multiple tunnels for each switch if that can be done (seems like it could be setup that way).
Also, need to try and configure the tunnels to be jumbo frames MTU 9000 (minus the GRE tunnel overhead of course). Have noticed that throughput is a little slower on the spoke LXC host nodes for example when installing packages from remote repositories so it seems to be a throughput issue and the likely cause is the smaller frame sizes due to the encapsulation of GRE tunnel packets.
So getting back to the main issue, these added networks would be used for example typically in my personal case for setting up Oracle 12c RDBMS ASM Flex clusters and these nets have to be able to talk across the GRE tunnel to other interfaces on that same network in the other "RAC" containers.
Additional networks need routing so that they can also use the sw1 tunnel rather than adding multiple tunnels for each switch if that can be done (seems like it could be setup that way).
Also, need to try and configure the tunnels to be jumbo frames MTU 9000 (minus the GRE tunnel overhead of course). Have noticed that throughput is a little slower on the spoke LXC host nodes for example when installing packages from remote repositories so it seems to be a throughput issue and the likely cause is the smaller frame sizes due to the encapsulation of GRE tunnel packets. So getting back to the main issue, these added networks would be used for example typically in my personal case for setting up Oracle 12c RDBMS ASM Flex clusters and these nets have to be able to talk across the GRE tunnel to other interfaces on that same network in the other "RAC" containers.