In retryable_writes_retry_conflict.js, we force a shard stepdown in some of the test cases, and quickly after attempt to execute a write on mongos. Retryable writes are not enabled, so if mongos hasn't yet received the new topology info, this write can fail with NotWritablePrimary e.g. We should ensure mongos knows the new primary before [attempting the writes|retryable_writes_retry_conflict.js].