The WT inMemory storage engine does not need to maintain snapshot history nor does it have a metric which MongoDB can use to gauge cache pressure and control the snapshot history window size. The window size should be kept at 0 for v4.2. inMemory members of a replica set cannot actually run transactions in v4.2, though they can replicate them.

Voting inMemory replica set members require the replica set config to set the writeConcernMajorityJournalDefault to false. Replica sets with such configuration are not currently allowed to run replica set / shared transactions.

An inMemory replica set with writeConcernMajorityJournalDefault set to true must be non-voting. Therefore, such a member cannot be primary. Transactions are only run against primary replica set members. So again, the inMemory node cannot directly run transactions, only replicate them as a secondary.

There is no need for inMemory nodes to maintain snapshot history.