Seems like this is a known issue in 8.6.0.4 and 8.6.0.5 at least. I solved the issue the hard way. I do it the hard way because its a LAB environment, take special care in production!!
- When i make the MM02 master and MM01 standby the issue was gone, when make MM01 master again the issue was back. In my situation.
- Make MM02 the master again.
- Stop the database synchronization.
- Change the vrrp priority so MM02 keep priority.
- write erase all MM01.
- Restore vrrp settings.
- enable the database synchronization where MM02 is master and synced to the new MM01.
- Make MM01 master again.
- Problem solved, without downtime or loosing licences.
Better contact TAC support if you have this issue in production.