We have chronic provisioning problems on Nebula and no documented process for troubleshooting. This has led to hours of wasted time.

A few examples:

• The 8/2/16 MTU change caused a number of problems for us, even though we were apparently on notification list
• We do not receive notifications of security blackholing servers
• A networking problem, possibly related to changes in zone policies, caused us to not be able to provision new clusters for testing for ~48 hours

We need to document a process for troubleshooting Nebula issues. The most straightforward approach seems to me to

1. Find a repeatable test case for the problem
2. File an issue with Nebula team
3. Document interactions and resolution

Most importantly, we need to not try and fix these problems until we've confirmed that nothing has changed in the OpenStack configuration.