Details
Description
If the config server steps down before it gets a failed moveChunk response from the donor shard, when the balancer thread gets recovered, it will load the failed migration and retry it. So the migration will not get aborted as the test expected.