Currently if replicaset.replica is not set and master is not available (after master auto-discovery it might easily happen), the read requests might fail. Even if there are replicas not equal to replicaset.replica nor to replicaset.master. It might be better to failover to a working replica right away without waiting for the failover fiber to catch up.
Currently if
replicaset.replicais not set and master is not available (after master auto-discovery it might easily happen), the read requests might fail. Even if there are replicas not equal toreplicaset.replicanor toreplicaset.master. It might be better to failover to a working replica right away without waiting for the failover fiber to catch up.