I am also periodically getting the same error (about once every 2 months). Replication is on a cluster and the only way I have been able to resolve the problem...