In the fsm workloads create_capped_collection.js and create_capped_collection_maxdocs.js, we check the number of documents after truncations and the actual documents. This is done with 2 queries and the number of documents is used to assert that the actual documents are correct.

The problem is if the stepdown thread is running and the primary steps down between the 2 queries the results may not be consistent since capped collections are maintained independently in the replica set.

The same assertion could be done with a single query, which would remove the window in which a stepdown could cause problems.