For your perusal, I've posted a DHCP agent patch that one of my colleagues developed for a customer who was seeing the DHCP agent appear to get behind with its handling of new VM ports.
I've just rebased this up to current master with no further testing, so it's quite unlikely to be exactly correct as currently posted. But you may find the approach interesting.
I'm going to dig out and report more history of exactly what we were observing at the time, so that we can evaluate whether it's really the same scenario as this bug.
For your perusal, I've posted a DHCP agent patch that one of my colleagues developed for a customer who was seeing the DHCP agent appear to get behind with its handling of new VM ports.
I've just rebased this up to current master with no further testing, so it's quite unlikely to be exactly correct as currently posted. But you may find the approach interesting.
I'm going to dig out and report more history of exactly what we were observing at the time, so that we can evaluate whether it's really the same scenario as this bug.