The evergreen/antithesis_image_build_and_push.sh script runs a sanity check to verify a test run by resmoke against a Docker Compose spawned MongoDB cluster will succeed. The current approach runs the same test file each time the Evergreen task executes which happens to correspond to the test with the smallest name lexicographically.

The antithesis_config_fuzzer_stress_concurrency_sharded_replication switching from running the agg_base.js FSM workload to the add_remove_shards_killop.js FSM workload (SERVER-81974) led to the discovery of a latent limitation for the test infrastructure which the Docker Compose setup relies on (SERVER-83368). Shuffling the test order would be one way to build up confidence over time that tests are capable of running without error under a Docker Compose setup similar to the one Antithesis uses. Other approaches increase the validate the test setup should also be considered by the team.