On Sat, Aug 25, 2018 at 01:35:17AM +0200, Stefan Schoerghofer wrote: > Hello, > > Today I've tried to use the following setup using openvswitch. > > > Three servers patched in a ring: > > ------------------- > | server 1 |--- > ------------------- | > | | > ------------------- | > | server 2 | | > ------------------- | > | | > ------------------- | > | server 3 |--- > ------------------- > > All ports within the ring are added to a bridge with rstp enabled. > Setup like this: > > ovs-vsctl add-br br0 -- set bridge br0 rstp_enable=true > ovs-vsctl add-port br0 ens4 vlan_mode=native-tagged > ovs-vsctl add-port br0 ens5 vlan_mode=native-tagged > ovs-vsctl add-port br0 ens6 vlan_mode=native-tagged # Not patched > # Adding a internal port > ovs-vsctl add-port br0 p1 tag=9 -- set interface p1 type=internal > ip addr add 10.0.0.1/24 dev p1 > ip link set p1 up > > After setting up (server2 has of course the ip 10.0.0.2, server3 > 10.0.0.3) the servers can not reach each other. > After restarting the openvswitch-switch service everything begin to > work as expected. > Before restarting it looks in tcpdumps like the traffic hits the bridge > but is not forwarded to the internal port.
Is this likely related to (R)STP? That is, if you disconnect one of the links and turn off (R)STP, do you still see the same behavior? _______________________________________________ discuss mailing list [email protected] https://mail.openvswitch.org/mailman/listinfo/ovs-discuss
