SERVER-116528 addressed a race in balancer-initiated force-jumbo migrations during shard draining. When the jumbo migration path uses an index scan, concurrent shard-key updates can cause the same document to be cloned twice, resulting in duplicate _id errors on the recipient shard and an aborted migration. That mitigation was intentionally limited to shard-draining migrations because repeated failures could prevent removeShard from making progress. Normal-balancing force-jumbo migrations were not addressed.

During normal data-imbalance balancing with "attemptToBalanceJumboChunks:true", the balancer can still invoke the force-jumbo path for large chunks, making it vulnerable to the same duplicate-key race. 

Two possible solutions:

• The balancer should not automatically attempt migrations of jumbo chunks.
• A more comprehensive fix would allow automatic jumbo migrations to proceed safely by entering the critical section before cloning, as manual forceJumbo migrations do.