Before starting a large ETL job, ETL operators can manually request a Recovery Log Checkpoint, instead of relying on the Sequencer automatic checkpoint process. The Recovery Log Checkpoint can also be set up at a time with low database activity so that in-flight transactions do not interfere with the checkpoint process. Similar to the automatic checkpoint process, the Sequencer must wait for all in-flight transactions to close. The default timeout is 90 seconds. If the transactions do not close, the Sequencer fails the checkpoint process and creates an error. The HaltTimeout parameter controls the default time-out value.
When the timeout setting is used, all new transactions started by clients are held until the timeout expires or the checkpoint occurs. This may cause delays in processing client requests.
SyntaxRECOVERY LOG CHECKPOINT [NOW | <timeout>];
The attempt does not wait. If there are any in-flight transactions, the checkpoint fails immediately.<timeout>
The attempt waits the specified timeout for in-flight transactions to complete. This overrides the timeout specified in the HaltTimeout parameter.
unityadmin -u admin -P admin -e 'recovery log checkpoint now'
unityadmin -u admin -P admin -e 'recovery log checkpoint 60'
unityadmin -u admin -P admin -e 'recovery log checkpoint'