On 6/2/25 1:33 PM, Felix Huettner via dev wrote: > Previously local ports that had an external_id:iface-id set would get a > ct-zone assigned on recompute, independent if there is a matching > southbound port_binding for them or if this port_binding can actually > reside on the local chassis. > However during incremental processing this was not the case and only > bound local ports would have their ct zone added or removed (as only > they are included in tracked_dp_bindings). > > To fix this we add a status enum to local_lports that defines if there > is a matching port_binding or not. > The ct_zones engine node no only considers local_lports that are backed > by some port_binding entry. > > This means we can also get rid of workarounds in testcases that where > introduced in 9f052bdb because of this problem. > > Acked-by: Ales Musil <amu...@redhat.com> > Signed-off-by: Felix Huettner <felix.huettner@stackit.cloud> > --- > v8->v9: addressed review comments > v6->v7: > * introduced > * addressed review comments >
Hi Felix, Ales, Thanks for the patch and review! Applied to main. CC: Han, Mark, Numan This is a bug fix but I'm a bit on the fence about backporting it because it depends on patch 1/3 and I'd prefer if we let that one soak a bit on the main branch. Any thoughts? Regards, Dumitru _______________________________________________ dev mailing list d...@openvswitch.org https://mail.openvswitch.org/mailman/listinfo/ovs-dev
