Comment 0 for bug 1874400

The revision_number is used to check if the message received at the agent is stale or not. However neutron doesn't consider a message as stale if the revision_number in NetworkCache is equal to the revision_number of the incoming payload. It is only considered stale if the revision_number in NetworkCache is greater than the revision_number in incoming payload

https://opendev.org/openstack/neutron/src/commit/10230683a2ce2f26279feaa34af4c0eccbfcb16c/neutron/agent/dhcp/agent.py#L852

Because of this lot more messages not considered as stale even though they are. This causes lot of "reload_allocations" and dhcp reloads and can cause bottlenecks in busy openstack environments.