Re: High level record lock

Charles (and others), you are of course correct. I definitely could not see part of the forest because of all the trees.

I dug into this further and found that the interactive job is doing:
SQL SELECT
Increment number
SQL UPDATE

Well DUH! No wonder this happened!

I need to replace the SQL logic with good old-fashioned:
CHAIN (locking the record)
Increment number
UPDATE

Thanks for not giving up until the light finally shined through.

~TA~



On 9/26/2018 4:02 PM, Charles Wilt wrote:

You're missing the point...

If you're using RPG RLA (CHAIN (or READ)/UPDATE)...

Then the record is locked unless your using the (N) extended to not lock.

The lock will be held till you do an UPDATE or another read...

Another RPG program will not be able to READ the record till the lock is
released.

Thus, what you are describing is not possible.

SQL is a different matter. If your SQL process is doing a SELECT INTO, it
will be able to read the record even if it's locked by another RPG program.

Your SQL process would need to be changed to use
DECLARE CURSOR
FETCH INTO
UPDATE WHERE CURRENT OF CURSOR

Now the SQL process will wait to read till the record is released.

Charles