If I'm not mistaking, the frozen buckets name will contain the epoch time (db_) the same way we have with the warm and cold buckets. therefore you could create a simple script which will check the earliest time of each bucket from the name and remove all the old one.
The frozen data will be stored in a NFS volume and i would like to know if there is a way to control the total disk space used by the frozen buckets.
No, the frozen buckets are considered as out of splunk, their size and volume are not monitored or controlled.
They are under the responsibility of your archive administrator to move them or rotate them to ensure disk space for the future frozen buckets,.