Task #1789
Task #1782: CN Replication components should be separated for scalability
Integrate replication queues and maps into Hazelcast processing cluster
100%
Description
The ReplicationService class was previously only connecting to a single Hazelcast cluster. It is now defined as ReplicationManager, and will connect to the storage cluster initiated by Metacat as a HazelcastClient and to the processing cluster initiated by the batch daemon as a HazelcastInstance cluster member. The queues and maps defined in "this configuration file":https://repository.dataone.org/software/cicore/trunk/cn/d1_replication/src/main/resources/hazelcast.xml should be integrated into the default configuration file for the processing cluster.
Specifically, add the hzReplicationTasks queue, the hzReplicationTasksMap map (that backs the queue), and the hzPendingReplicationTasks map. The hzSystemMetadata map will be added to the Metacat storage cluster, so it doesn't need to be transferred.
History
#1 Updated by Chris Jones over 13 years ago
- Subject changed from Integrate replication queus and maps into Hazelcast processing cluster to Integrate replication queues and maps into Hazelcast processing cluster
#2 Updated by Robert Waltz about 13 years ago
- Status changed from New to Closed
