I’d like to cherry-pick this change into support/1.13: There is a flaw in the code that handles a server being forced out of the cluster. The flaw keeps the server from shutting down and leaves the server in a hung state. The PR adds error handling to two methods, one in the Cache’s XML-generation method and the other in a Membership class to ensure that tear-down of the cache is performed.
https://github.com/apache/geode/pull/5490