DHCP Timeouts

Seeing issues where the vif isn't getting successfully plugged for dnsmasq, which can cause a metadata failure due to no DHCP leases being issued. I do know that restarting the neutron-dhcp-agent service causes approximately a 5-minute service interruption as addresses are issued/reissued, which is far too long. We may be at a point where we have to do some performance tuning.

This seems helpful.

https://www.mirantis.com/blog/improving-dhcp-performance-openstack/

Who wants this ?