Distributed Cache in MapReduce can be updated by replacing the file with the new one and changing the pointer location to point to new location and restart the MapReduce job or by appending the values in Distributed cache and restarting the job.
Note: We cannot update the Distributed Cache when the MapReduce job is running. It will become a race between the two operation in which both will lose.
We have to restart the job and submit another DistributedCache data. Distributedchache is not persistent between jobs.