Hi, I've ended up solving this problem. The interface state was not being changed because I was handling a GRE tunnel in OpenvSwitch and shutting down the physical interface down (the state of the physical interface was being set as down, but the GRE interface state was still up).
I've made some new experiments without using GRE tunneling, using a physical interface instead. With that scenario it worked, as the interface state was modified and the fast-failover mechanism sent the packets through the other bucket in the group. So right now my question gets more specific. Is there any method for getting fast failover to work when using GRE interfaces in OVS? How can a GRE interface state get flagged as down? Thank you. Best regards, Ricardo Santos On 21 June 2014 18:41, Ricardo Santos <rrsan...@student.dei.uc.pt> wrote: > Hello, > > I have some questions about the fast failover group option when using > OpenFlow 1.3 in Open vSwitch 2.1.2. How does Open vSwitch monitor the state > of the present buckets in a group? > > Supposing that I have one group created by the following command: > >> ovs-ofctl -O OpenFlow13 add-group br0 >> group_id=123,type=fast_failover,bucket=watch_port:1,output:1,bucket=watch_port:2,output:2 > > > and primarily the packets are being sent through port 1, how can I make > that port being deliberately not alive, so the traffic can be sent to port > 2? I've tried through an *ifconfig eth0 down* command (affecting the > interface to which port 1 is mapped), but the interface in OVS doesn't > change it state. > > Thanks in advance. > > Best regards, > Ricardo Santos >
_______________________________________________ discuss mailing list discuss@openvswitch.org http://openvswitch.org/mailman/listinfo/discuss
