Synchronization Phase
In this phase, the --jnlsync or --jnlsync_only process must be run at regular intervals through a cron job. The steps for the --jnlsync and --jnlysync_only processes are as follows:
1. If the source has taken a checkpoint since the previous jnlsync operation and --jnlsync is used, copy the source checkpoint to the target(s) and roll forward.
Note:
- Rolling forward a new checkpoint requires the target installation(s) to be shut down.
- If --jnlsync_only is used, new checkpoints are not copied.
2. Close the current journal file and create a new one.
3. Copy the new journal files from the source to target(s).
4. Apply the journaled transactions to the target database(s).
Target node databases are in the INCR_RFP state. In this state, user connections are accepted, but only SELECT operations are allowed. Update operations are not allowed and will fail. Any target database connections can cause the --jnlsync step to fail with appropriate error messages. This is a soft failure because subsequent invocations will attempt to roll forward all the unprocessed journals. The synchronization step on the target(s) will connect with the source if there is a window with no user connections. If the source fails, the target will be caught up when it is designated as the new source.
Last modified date: 09/11/2026