Reading through the implementation of shardColl(), the resulting behavior of the operation (in terms of where data ends up) is not clear.

If I provide "collName" and "key", it's unclear to me if I should expect the data to actually be distributed across shard or just reside on the primary shard since I don't provide any split or move logic. This came up in code review with charlie.swanson@mongodb.com when testing SERVER-106933. I did some manual checks to confirm the 3 documents do get split across the 2 shards, but it's not clear to us why. Is it the balancer? Is the helper finding a split point for us?

We've had a history of bugs on query because we think we're testing a sharded collection but the test doesn't actually distribute documents, or vice versa. It'd be much appreciated if ShardingTest was more intuitive to work with in this case