Multiple instances of CDAP can run on a single HBase cluster. Each instance of CDAP will get its own co-processor for its tables. And each co-processor will read the right transaction snapshot file for the instance.
However, the log message emitted after a snapshot reload does not contain the instance of CDAP for which the snapshot was loaded for. It just says Transaction state reloaded with snapshot from 1500220336000. This makes it difficult to figure out the right prune timestamp when performing manual pruning of invalid list.