There is only one log per database for the cluster, and it is logically partitioned. A log record marker for the Cluster Edition contains the instance id, which it used internally to scan log records logged by specific instances, especially during failover recovery. A checkpoint log record always has an instance id with a value of zero, so it is always included in the log scan, regardless of the instance id specified in the log scan.
