It looks like our test setup is having some problems. I'll look into it. For now, ignore this...
Best Regards, Daniele Il 28 gen 2016 5:11 PM, "Daniele Venturino" <daniele.ventur...@m3s.it> ha scritto: > Hi, > I was running some tests on Mininet and Open vSwitch and I noticed > something weird. I'm using few bridges configured with RSTP enabled and no > controller. > I then generate some traffic and I'm able to see the flows installed on > each bridge through the command > > *ovs-ofctl dump-flows br0* > > If I disable an active port (from which the traffic was flowing), I can > still see the flows of that port, while I'd expect them to be moved > (updated) on a new active path connecting the two ping endpoints. > > If I repeat the same test, but setting the switches in normal mode with > > *ovs-ofctl add-flow br0 action=NORMAL* > > I can see the forwarding tables though the command: > > *ovs-appctl fdb/show br0* > > and if I disable a network interface where a MAC was learned, its entries > are correctly removed from RSTP, with this piece of code in > ofproto/ofproto-dpif.c: > > * while ((ofport = rstp_check_and_reset_fdb_flush(ofproto->rstp, &rp))) {* > * if (!rstp_shift_root_learned_address(ofproto->rstp)* > * || rstp_get_old_root_aux(ofproto->rstp) != ofport) {* > * bundle_flush_macs(ofport->bundle, false);* > * }* > * }* > > Is it ok for the flows not to be deleted when a port becomes inactive? Am > I missing something? > > Best Regards, > Daniele > > > > _______________________________________________ dev mailing list dev@openvswitch.org http://openvswitch.org/mailman/listinfo/dev