Re: Data area LOCK timeout

On 26-May-2016 12:04 -0500, Bradley Stone wrote:

I'm having a hard time finding an answer to this..

Let's say PGMA reads in data area DTAARA1 with *LOCK, updates the
data area value, then uses OUT to update the data area (and unlock
it).

PGMB also does the same thing. Let's say they happen to run at the
same time (but PGMA got there first)... how long does PGMB wait until
it issues an error saying it can't lock the data area?

The update is very quick.. it's really just incrementing a counter
value stored in the data area and returning the value.

My SWAG, the Default Wait Time (DFTWAIT) established for the job [from the Class (CLS) object or the value on DFTWAIT set with Change Job (CHGJOB)].

Would seem easy enough to gather empirical data; easiest if instead of stopping PGMA before the unlock, that is just replaced with Allocate Object (ALCOBJ) of the Data Area using the *EXCL Lock State. Call PGMB with the current DFTWAIT, then after timeout for that invocation, change the job to a much shorter DFTWAIT value and make a new invocation of PGMB. If the timeout for the second invocation is similarly much shorter than the prior invocation, then the lock by the RPG run-time uses [a zero-value as the wait time in the MI LOCK template for the Lock Object (LOCK) MI instruction, which means] the DFTWAIT attribute from the job for the Wait Time.

And without actually reviewing any, I found these past topics which might be related [I apparently replied to the first topic]:

[http://archive.midrange.com/rpg400-l/201111/threads.html#00104]
Subject: Wait time on a *DTAARA...

[http://archive.midrange.com/rpg400-l/200305/threads.html#00728]
Subject: Dataarea lock problem

[http://archive.midrange.com/midrange-l/201410/threads.html#00887]
Subject: adjust wait time to acquire an object