Closed youngm closed 8 years ago
The new quorum code appears to work great. If quorum is not met then the memcache server is shut down until quorum is available again.
This code doesn't appear to get in the way of the new drain/post-start scripts either.
I do some custom quorum stuff currently in to ensure the cluster doesn't try to handle requests if not enough members exist in the cluster. It appears hazelcast added formal support for quorums. We should investigate using that instead.