Currently, in sharded clusters, we run validate on the config server and the shard servers. However, we don't actually expect that the shard server validate should ever fail if the config server does, and if it does, it will cause an infinite loop. Therefore, this second validation step should be removed.