We had an issue after upgrading Gluster nodes to 6.6, we had memory leak in
gluster self heal daemon which cause it to do infinite heal and ate all of
the nodes RAM "128GB", we had to downgrade to 6.5 to solve the problem.
I think it is related to the changes made to the SHD code in version 6.6 (
1760706 <
https://bugzilla.redhat.com/1760706>)
I might find the time and try to bug report the issue.
Anyway, glad you solved your issue.