Writing to TS

[stansaraczewski]

I'm debugging an assembler program that is abending on a item error after a read ts.

The situation is complicated by the user not being able to reproduce the error on demand... happens some of the time.

Another issue is that no diagnostic tools are available (other than CEDF of course).

The code is part of a very large system that is perhaps nearly 30 years old. It has many macros that produce the exec cics statements. The one I am concerned about is the write ts that creates the ts items that are read on re-entry... it has a nohandle specified, and is dealing with aux... the vsam temp storage dataset.

I've been away from CICS applications work for a number of years, but do recall from those 'old days' that specifying 'aux' wasn't good, and of course the nohandle will ignore errors.

Does anyone have any thoughts that they'd like to share ?

 

[stevexff]

The idea of specifying NOHANDLE is that you are supposed to check DFHRESP and sometimes DFHRESP2 afterwards and act accordingly. If it's not doing that, then perhaps a quick & dirty way would be to put in a check that will abend the transaction so you can get a dump out and see what's going on.

If he's doing a TS write to AUX then it's possible that the VSAM file is filling up - you could check the CICS stats to see if this is the case. From memory, there are things you can do to force TS to go to MAIN without changing the code via resource definitions

AUXILIARY
specifies that the temporary storage queue is on a direct access storage device
in auxiliary storage. This is the default value for the first write.
This option is ignored:
* for an existing queue,
* if a TSMODEL resource definition with a matching prefix is installed in the
system.

but obviously this depends on how much stuff you write to TS and how much memory you have to spare...

Steve

[small]"Every program can be reduced by one instruction, and every program has at least one bug. Therefore, any program can be reduced to one instruction which doesn't work." (Object:erlDesignPatterns)[/small]

 

[stansaraczewski]

Thanks for the reply Steve. Yes, I did look into it a bit further to see that indeed DFHRESP is being examined. I am getting a dump, but it's reformatted by ABEND-AID and the passage of years since I have last seen a CICS transaction dump is making it difficult.

Just got yanked off the project anyway -