In parallel of the Rodolfo fix https://review.opendev.org/c/openstack/neutron/+/806246 maybe we can try to fix race condition between applying deferred flows and removing conntrack entries?
That would fix an issue with having a bad conntrack entries for invalid connections created during openflow batches apply?
Something like: https://review.opendev.org/c/openstack/neutron/+/837726
In parallel of the Rodolfo fix https:/ /review. opendev. org/c/openstack /neutron/ +/806246
maybe we can try to fix race condition between applying deferred flows and
removing conntrack entries?
That would fix an issue with having a bad conntrack entries for
invalid connections created during openflow batches apply?
Something like: https:/ /review. opendev. org/c/openstack /neutron/ +/837726