/nodeStatuses intermittently reporting unhealthy status

We are running a 6 node cluster of Couchbase 2.2 and it is being monitored by curling /nodeStatuses every minute. About ten times a day we are seeing an unhealthy status for a single curl and then it goes back to healthy. I have looked through the logs for the time period a node was unhealthy and can’t find anything significant. Any suggestions on what could be happening?