Infrastructure

Science Metadata must also be replicated across the CNs, and the current Metacat implementation provides a serial queue to process replication events one by one. To facilitate parallel replication across CNs, use the Hazelcast ExecutorService to submit CNReplicationTasks to each of the CNs in the cluster. The PID should be locked until the task is finished for the science metadata object. Although we can submit to a Set of hazelcast members at once with one task, it may be better to iterate over the set and submit the task to each member in order to individually manage locks and failures for each CN.