Parallel Checkpoint and Restart
All instances must call the Checkpoint function in unison. While only the master instance will issue the actual checkpoint to the database, all the slave instances must call the Checkpoint function at the same time to ensure that all rows loaded using the PutRow function or the PutBuffer function are sent to the database prior to the Checkpoint.
Checkpoint data for each instance must be copied and stored separately. Each instance must use its own checkpoint data when restarting. For more information, see “Add Checkpoint and Restart” on page 33.