If multiple Data Mover servers are installed in your environment, you can use the Data Mover synchronization service to create a backup system that can be used as a temporary master system in case of daemon failure on the original master system.
- Master server
- Refers to the Data Mover server on which the daemon is running.
- Slave server
- Refers to one or more Data Mover servers on which the daemon is not running. Slave servers can be used as a primary system in case of failure on the master server.
Replication begins for the first time when the synchronization service running on the server that is used for the slave repository connects through a TCP/IP socket to the system that is used for the master repository. When the synchronization service starts, updates to the master repository are replicated automatically on the slave repository.
As part of the failover functionality, when the slave server is unavailable (offline, down, restarting), the master repository continues to log updates. When a connection is re-established with the slave server, all interim updates from the master repository that occurred when the slave server was unavailable are applied to the slave repository.
If the slave is down too long, interim updates will accumulate and may cause a memory issue on the master synchronization service. When a slave node connects to the master, the master synchronization services checks to see if there are too many updates for the slave; if so, the master synchronization service disconnects the slave node and generates the exception "Slave has xxxx bytes sqls to be processed since last disconnection, which exceeds current available memory on the master sync service. Slave node will be disconnected. Please run backup and restore to sync the master and slave nodes." As indicated in the message, run backup and restore to sync the master. For more information, see Synchronizing the Master and Slave Repositories.
File Name | Description |
---|---|
/etc/opt/teradata/datamover/sync.properties | Settings that the Data Mover Replication Service uses for synchronizing master with slave repositories. |
/opt/teradata/datamover/sync/nn.nn/DMReplication.jar | Executable binary file used by the synchronization services. |
/opt/teradata/datamover/sync/nn.nn/dmsync | Script for starting the synchronization service. |
Usage Notes
You can use host name or IP address as values for the master.host and jobstore.host in sync.properties for the master and slave servers. When configuring the servers, use only all host names or all IP addresses. Do not mix host names and IP addresses.
Scenario | Result |
---|---|
The sync.properties file is not located at /etc/opt/teradata/datamover | The synchronization service reports an error. |
The dmsync or DMReplication.jar file is not located at /opt/teradata/datamover/daemon/nn.nn | The synchronization service reports an error. |
The slave synchronization service is started before the master synchronization service | The synchronization service reports an error. |
The master and slave synchronization services are started on the same server | The synchronization service reports an error. |
The master repository is unavailable (offline, down, restarting) | The synchronization service stops all connected slave repositories. |
The master repository is restarted after a period of unavailability | The synchronization service reports an error. The master sync service must be restarted by running /opt/teradata/datamover/sync/nn.nn/dmsync start, where nn.nn in the path refers to the major and minor version numbers of Data Mover. Previously connected slave services must be restarted individually by running /opt/teradata/datamover/sync/nn.nn/dmsync start for each one. |