Report Specifications Archive Definition exit - on z/OS from Windows?

[HeatherB]

We are running CMOD v9.0.0.3 on z/OS (and moving to v 9.5.0.3).  We currently utilize the Report Specifications Archive Definition exit for ARSLOAD in order to identify which storage node is to be used for loading the data.  Our Application teams are looking to run more of their processes on a Windows Server.  From their, they would like to load to the CMOD servers running on z/OS.  We do this on occasion today with smaller documents, and the default storage node is used.  Going forward, is there any way to trigger off the Report Specifications Archive exit, on z/OS, when running the load from the Windows Server?  Considering the volume of data, utilizing the default storage node is not going to be practical. 

Thank you for any insight provided.

Heather

[ewirtz]

Hi Heather,
if you are running arsload on the windows server against an archive on the mainframe, you can provide an own Report Specifications Archive Definition exit on the Windows Server that overrides the default storage node.
regards
Egon

[Justin Derrick]

I've never heard of this exit -- can someone link to some documentation for it?

[HeatherB]

Egon,

Thank you for pointing that out.  I see now where the arsuupdt resides in the exits directory.  My issue is that our exit was customized by Lab services many years ago.  The version we use is a COBOL version, on the mainframe, which is accessing data in a DB2 table for determining the storage node.  I'll need to look into that some more.

Thank you,
Heather


JBNC - Some information on the exit can be found here, on page 269: http://www.redbooks.ibm.com/redbooks/pdfs/sg246915.pdf.  Also, "For more information about the report specifications archive definition exit routines, see Chapter 40, “Report Specifications archive definition exit”, in the Content Manager OnDemand for z/OS Configuration Guide, SC19-3363."

[ewirtz]

Hi Heather,
I think on Windows it is the best choice to use C for the exit. You can access the mainframe table  or a local copy of the table.
regards
Egon

[Nolan]

I am in the same situation as Heather.  We have our COBOL exit on Z/OS and now need to run PDF Indexer and loading from the Windows server.

1. Does any one have the steps to follow to set up the server.
2. Do I need to install DB/2 on the windows server to get to the Z/OS OnDemand tables?
3. Anyone have a sample of the code that is better than what IBM posts 

#pragma export(UPDTEXIT)
#include "arscsxit.h"

int UPDTEXIT( ArsCSXitUpdtExit  updt )

Thank you

[Nolan]

I have moved forward a little bit but not much.  Was hoping HeatherB would comment.

Setting up the DB/2 connection from the Windows server using the .NET toolset is proving to be a challange. 

Q.  When compiling the arsuupdt.c are you able to compile in VS 2013 or do you need resort to line commands?  ie DB2 PREP and CL

[HeatherB]

Hi Nolan,

Have not been back in the ODUG boards in awhile.

Regarding the exit running on the server, we have not implemented anything to date.  We have, however, done some brainstorming of different options and had a few conversations with IBM.  One of our latest solutions would involve IBM lab services rewriting the existing z/OS exit to run on Linux.  With that, the storage node information required still resides in DB2 on z/OS.  We would need to use DB2 connect to access the information back on the mainframe. 

From our conversations a few questions/suggestions have come up from our Application teams regarding storage nodes and default storage nodes.  This brought me back to the support forums for further discussion.  I'll be posting that in a separate thread, however.

Good luck with your implementation. 
Heather


 

[Nolan]

I also gave up on the solution to talk to DB/2 from the C exit.  I think it is possible but not well documented and would require the services team or other developers involved.   

Thinking outside of the box and also keeping in mind that I didn't want to impact our current process, we have hundreds of JCLs already in place that I didn't want to update.
I created the following to replace our existing proc.

1. Using the application ID passed from the JCL, call a REXX program to issue the same query to DB/2 as the report specification Exit
2. The REXX creates our temporary sysin for Universal command to copy the PDF file from Z/OS down to the new Windows PDF Server
3. The copy of the PDF file down to the server with a new name of APPNAME.AGNAME.STORAGE_NODE.PDF

Since it is all done in the PROC, all I have to do is distribute the new proc and none of the JCL's need to be updated.

4. Using the Windows Report specification C exit, I parse out the file name setting the APPNAME, AGNAME and STORAGE Node

The local ARSLOAD service picks up the file and processes it perfectly back in to Z/OS. 

PM me or Ed if you want to have a detailed discussion.

Cheers

EDIT:

I realized that using the Windows service to load the PDF documents would then require some other process to capture success or failure of the loads.  I have disabled the Windows service and now have added another step to the proc.  The last step of the proc issues the ARSLOAD command on the Windows server and the return code of ARSLOAD is passed back to the executing JCL.