Major - P3 The `migration_sets_fromMigrate_flag.js` test occasionally fails in the continuous stepdown suite because the config server balancer steps down and cases moveChunk to run one time too many, which causes more 'i' entries to be present in the recipient's oplog.
Specifically, from the logs it can be seen that the migration, which fails due to 'failMigrationLeaveOrphans' happening twice due to the balancer recovery:
d23264| 2017-12-18T22:20:05.678+0000 W COMMAND [conn29] failpoint: failMigrationLeaveOrphans set to: { mode: 1, data: {} }
...
d23264| 2017-12-18T22:20:08.319+0000 I SHARDING [migrateThread] failing migration after cloning 1 docs due to failMigrationLeaveOrphans failpoint
d23261| 2017-12-18T22:20:16.026+0000 W SHARDING [conn26] Chunk move failed :: caused by :: OperationFailed: Data transfer error: failing migration after cloning 1 docs due to failMigrationLeaveOrphans failpoint
...
d23264| 2017-12-18T22:20:16.210+0000 I SHARDING [migrateThread] failing migration after cloning 1 docs due to failMigrationLeaveOrphans failpoint
This test should be blacklisted from the continuous stepdown suite.