We have seen issues downstream where calling dhcp_release didn't cause the lease to be removed from the leases files used by dnsmasq. There are a couple of scenarios where this could happen:
1. The packet is simply lost, as it is UDP, even though it's being looped-back
2. dnsmasq is reloading, so there is noone to receive it when it arrives
For that reason we should make this more robust. A couple of possible solutions are:
1. Send the release more than once in succession. It's probably OK to just send some small number of packets for each lease we want to release, it would easily increase the odds that one makes it through.
2. Check the leases file to make sure dnsmasq processed the release, and re-send for those addresses that were missed. This method is slightly more complicated, but it should also increase the odds, and should do it with fewer packets being generated.
Each option has some overhead, but since the option is not releasing the lease, it's worth it.
I have proposed option #2 already at https://review.openstack.org/#/c/560703/ but wanted to make sure to get feedback on other proposals that might also solve the issue.
Reviewed: https:/
Committed: https:/
Submitter: Zuul
Branch: master
commit fab032b426e1fa8
Author: Brian Haley <email address hidden>
Date: Wed Apr 11 21:22:23 2018 -0400
Retry dhcp_release on failures
Sometimes calls to dhcp_release(6) do not result in removal
of a lease from the leases file, for example, when the release
packet is not received by dnsmasq. Trying more than once is
recommended in this case.
Instead of blindly trying some number of times, we monitor
the lease file contents, and retry the dhcp_release(6) call
when an entry still remains. This is possible since
dhcp_release(6) is being run from the DHCP server itself.
We try three times and wait 0.3 seconds between tries.
We also now check for any stale leases in the leases file
that are unknown to neutron, also trying to remove them.
Change-Id: Ic1864f7efbc94d
Closes-bug: #1764481