What do you mean by Splunk? I believe are you referring to the Splunk Indexers, the answer is maybe. This all depends on your input.
If you are monitoring files the answer is no since the UF will pick up where it left off last. Though that does not protect against in-flight data loss. You can protect against in-flight data lost by using indexer acknowledgment.
If you data is streamed such as raw TCP data or Syslog data, yes you could lose data. There are ways to help protect against that. By using indexer acknowledgement and increasing queueSize on both your input.conf and output.conf files.
If you have multiple indexer you can use the UFs built in Auto Load Balancing feature.
Here is a good post on HA: splunk-disaster-recovery
