An FLE findAndModify command sends an additional command (getQueryableEncryptionCountInfo) to the shard prior to the findAndModify. This means that the findAndModify is the second statement in the transaction rather than the first and thus we cannot retry StaleConfig errors internally.

The find_and_modify.js test is running FLE findAndModify within a transaction and expecting success, but with a stale mongoS this may require a driver (or in our case a test) retry of the transaction.