Journaling considerations with DB2® Multisystem
Although the Start Journal Physical File (STRJRNPF) and End Journal Physical File (ENDJRNPF) commands are distributed to other systems, the actual journaling takes place on each system independently and to each system's own journal receiver.
As an example, you have two systems (A and B) on which you have a distributed file. You need to create a journal and a receiver on both system A and system B, and the journal name and library must be the same on both systems. When you issue the STRJRNPF command, the command is distributed to both systems and journaling starts on both systems. However, the journal receiver on system A contains only the data for changes that occur to the piece of the file that resides on system A. The journal receiver on system B contains only the data for changes that occur to the piece of the file that resides on system B.
This affects your save and restore strategy as well as your backup strategy; for example:
- After you issue the STRJRNPF command, you should save the database file from each of the systems in the file's node group.
- You need to practice standard journal management on each of the systems. You need to change and to save the journal receivers appropriately, so that you can manage the disk space usage for each system. Or, you can take advantage of the system change-journal management support.
Just the names of the journal must be the same on each of the systems; the attributes do not. Therefore, for example, you can specify a different journal receiver threshold value on the different systems, reflecting the available disk space on each of those systems.
- If you do need to recover a piece of a distributed database file, you only need to use the journal receiver from the system where the piece of the distributed file resided. From that journal receiver, you apply the journaled changes using the Apply Journaled Changes (APYJRNCHG) command or remove the journaled changes using the Remove Journaled Changes (RMVJRNCHG) command.
- You cannot use the journal receiver from one system to apply or remove the journaled changes to a piece of the file on another system. This is because each piece of the file on each system has its own unique journal identifier (JID).
Parent topic:
CL commands: Affecting all the pieces of a distributed file with DB2 Multisystem
Related concepts
System activities after the distributed file is created
CL commands: Affecting only local pieces of a distributed file with DB2 Multisystem