BMP Program and Checkpointing

sandiptcs · New User Joined: 29 Nov 2006 Posts: 19 Location: amsterdam

Is it mandatory that a BMP program using IMS and DB2 databases be coded for checkpointing

I have a BMP program which has a checkpoint at the beginning and at the end of the program
The reason , it says, a checkpointing is needed at the end to release the buffers. Any idea what is buffer is all about ??

The checkpointing in the program is handled like this

Initial section
Move spaces to restart work-area
Restart call
If normal start
checkpoint call
else
abend

Close section (at the end of the program)

Final checkpoint call

Bitneuker · Posted: Mon Jan 15, 2007 7:32 pm

Hi,

I don't see the use of checkpointing like you told. Click to learn about Checkpoint/Restart.

Nighthawk750 · New User Joined: 04 Nov 2006 Posts: 20 Location: Amsterdam

The reason why the program first checks to see if there is checkpoint data present, is that aparently no restarts are allowed. You see that when the abend is triggered.
In test-environment you can see in AES the checkpoint data (if you do not know how to use AES, please check with someone who does). You can then remove the checkpoint data and start again.
Should it happen in production environment, the operator present knows how to handle it.

The final checkpoint serves as an commit (eg. the database entries are made permanent) and yes, the IMS buffers will be emptied. During processing IMS uses its own buffers. While running the program, those buffers will be filled with data. It is data from the actions you perform on the database. When handling large amounts of data those buffers will eventually get at their max. So in that case an checkpoint is needed.

But your program aparently does not need those checkpoints. The amount of data processed is probably not that much.

sandiptcs · New User Joined: 29 Nov 2006 Posts: 19 Location: amsterdam

Nighthawk and Bitneukar
Thanks for your assistance.

The situation is a lot clear now. But I cant still understand the following comment you have made

Bitneuker · Posted: Wed Jan 17, 2007 7:34 pm

Look like an Amsterdam mini-meeting to me

Long ago I worked at NMB (now ING) where someone had a brainwave that all BMP's must be equiped with checkpoint/restart; even if it had no use. Perhaps this program has been subject to a similar brainwave. If it doesn't do any pain just leave it as it is. As explained to you checkpoint/restart for this thing is not necessary; either leave it useless or remove it.

Nighthawk750 · New User Joined: 04 Nov 2006 Posts: 20 Location: Amsterdam

@Sandip: If a program goes down, checkpoint data is kept in some kind of dataset / database. Even if the only checkpoint was the initial one.

When the program makes the restart call, in normal situations, there should be no checkpointdata for this program in the checkpoint database.
If there is, in this case, that is considered as a problem.
Aparently the one who designed the program wanted it to work like this. Initial chkp is only needed when IMSupdates are done during the program.

@Bitneuker: yeah, looks indeed like an amsterdam mini-meeting