Details
Major Description
Had a few customers refer to this page (http://www.couchbase.com/docs/couchbase-manual-2.0/couchbase-admin-tasks-failover-choosing.html#couchbase-admin-tasks-failover-automatic-considerations) and say that they fear our auto-failover because it can cause a split-brain.
The text that implies this is: If you have a network partition across the nodes in a Couchbase cluster, automatic failover would lead to nodes on both sides of the partition to automatically failover.
That certainly does sound like it would cause a split-brain. However, this is out of context because this text is supposed to indicate that "if we didn't do anything special, there would be a problem" but in fact, there isn't because we do prevent this case specifically.
I'm not sure I have the perfect solution at the moment, but this area probably needs a bit of reworking or cross-linking or simply a note being placed to make it clear that Couchbase auto-failover will NEVER cause a split-brain situation. Perhaps even a specific section about "Couchbase and split-brain" since it is a very important topic in the world of distributed databases.
As a stop-gap, could we put a quick note on this page (or just make a small edit to the text) to indicate this? And then setup a task to rework the discussion around autofailover...it was written by yours truly quite a long time ago.