In a sharded cluster, from which node does one stop the balancer process before initiating backup?

Question

Options

• AAny node
• Breplicaset primary node
• Cconfig server node
• Dmongos node

• A
  8% (2)
• B
  15% (4)
• C
  8% (2)
• D
  69% (18)

Explanation

Stopping the balancer through a mongos node is correct because mongos is the cluster's query router and the process responsible for coordinating chunk migrations between shards - it is the authoritative interface for balancer management commands like sh.stopBalancer(). Option A is wrong because the balancer cannot be stopped from "any" node; it requires the specific process that controls it. Option B is wrong because a replica set primary handles replication within a shard, not cluster-wide balancing operations. Option C is wrong because config servers store cluster metadata (including chunk distribution info), but they do not expose the interface to control the balancer process itself.

Memory tip: Think "mongos = router = traffic controller" - just as you'd tell a traffic controller to pause before doing road work (backup), you tell the mongos to stop moving data (balancing) before taking a consistent snapshot.

No community discussion yet for this question.