--Adjust exit_logging.js to skip the dbHashes check during shard stop. This ensures a focus on crash logging testing and prevents interference in test results. Additionally, note that crashed mongos instances may leave idle transactions on shards/config servers, causing resource locks and potential hang-ups in the dbHash check process. –

This test intentionally crashes the mongos and can make the transaction started by the ClusterServerParameterRefresher job become idle. In default, the transaction life time is 24 hours. In this case, the shard test stop operation will wait on `fsync` operation because this transaction hold for global resource lock for long time and cause the jstest timeout. We need to reduce it to less (eg, 10 seconds) to make this transaction to be quickly aborted by mongod itself.