On Error 3145, wait interval before RETRY

, ,
1

Hi friends,

Some times i have the 3145 error, and the code bollow are used to RETRY at least 100 times

ON ERROR
IF *ERROR-NR EQ 3145
ADD 1 TO #CONT-RETRY
IF #CONT-RETRY LE 100
RETRY
END-IF
END-IF

But, spending some times of search ive found a new code like ‘CMROLL’.

The ask is: can i use the code bellow with my programs to RETRY after 1 second?

ON ERROR
IF *ERROR-NR EQ 3145 AND #CONT-RETRY LE 5
ADD 1 TO #CONT-RETRY
CALL ‘CMROLL’ 1
RETRY
ELSE
ESCAPE ROUTINE
END-IF
END-ERROR

Its all right with this? This can degrade performance on other applications?

Tanks a lot and best regards.

2

The use of CMROLL is fine.

You are reseting #CONT-RETRY somewhere, right?

Basically CMROLL takes the object out of the thread; so it does not hurt other programs

steve

3

Which is a no-brainer on OpenSystems as there are no Natural threads.

4

But SAG generally frowns upon our use of its CM modules (entry points into the nuke). User exits are recommended instead. In this case, USR2027N.

5

USR2027N + RETRY is OK.

As an alternative you can use the parameter WH=ON. Especially if we’re talking about batch programs here…

http://techcommunity.softwareag.com/ecosystem/documentation/natural/nat6311win/parms/wh.htm

6

Right, but that’s just a yes/no switch which doesn’t let you decide on how
long you want to wait, it’s infinite (more or less, depending on timeout
settings).

7

You’re right. This could be a disadvantage.

An advantage is (from my point of view): with adaopr di=cq you can see the user’s waiting for a record to be released… BTW: Is this a new behaviour of adabas 6?

8

Nope, checked the documentation back as far as v3 of Adabas,
and the “Waiting for Isn ” status is already documented
there :wink:

9

Thanks!

But IIRC in Adabas 3.3@Solaris WH=ON resulted in an new adabas call every second…