one node has vbucket flapping active/inactive during rebalance

During testing of a client and during rebalance, we happened to notice one vbucket on one node (ip ending in 212) flapping out of active and back. At least, that's what the REST interface and web console were showing.

Scenario:
4 nodes

Steps:
1. Establish 4 node cluster
2. Start rebalance out of two of the four nodes

Observed behavior: vbuckets migrating from the two nodes to one of the two, and the other node with vbuckets flapping back and forth between 256 and 257 during the beginning of the rebalance.

Expected behavior: I would think that we'd be rebalancing to both, but even if it's just going to one, I'd expect the other node to not have temporary on-and-off of another vbucket.