USD025SS Failing every day with CC 00012 ** CRITICAL PATH ***

[Md.Imtiaz]

Hi,

We have a Job USD025SS, which abends with CC 00012. Upon checking upon it shows that one of the VSAM dataset is being used. When we try to find the job or user accessing the dataset, we can't find it but when we try to exeute the Job from abended step, it works. What can we do to mitigate it?

[Md.Imtiaz]

There is another interesting thing the VSAM in question does not have a INDEX part:

Code: Select all

	  	 Menu 	 Options 	 View 	 Utilities 	 Compilers 	 Help 	                             
 	------------------------------------------------------------------------------ 
 	DSLIST - Data Sets Matching V.PR50.VSAMDT                           Row 1 of 2 
 	Command ===> 	                                                 	Scroll ===> 	CSR  
 	                                                                               
 	Command - Enter "/" to select action             	     Message          	 Volume 
 	-------------------------------------------------------------------------------
 	         V.PR50.VSAMDT                                	                 	 *VSAM* 
 	         V.PR50.VSAMDT.DATA                           	                 	 PVS127+
 	***************************** End of Data Set list ****************************
 	                                                                               
                                                                                
                                                                                
                                                                                
                                                                                
                                                                                
                                                                                
                                                                                
                                                                                
                                                                                
                                                                                
                                                                                
                                                                                
 	  	*DSLIST  	-DSLIST  	                                                           

*DSLIST -DSLIST

[Robert Sample]

There is another interesting thing the VSAM in question does not have a INDEX part:

This is not actually an interesting thing. There are different types of VSAM data sets (ESDS, RRDS, KSDS, LDS) and an ESDS in particular does NOT have an index component. So you've probably got a VSAM ESDS data set you're looking at.

it shows that one of the VSAM dataset is being used.

What are the SHAREOPTIONS for the data set? Is the data set JCL reference DISP=OLD or DISP=SHR? These questions could indicate where your problem is.

[Md.Imtiaz]

What are the SHAREOPTIONS for the data set?

How do I identify about that?

DISP=SHR is used for this file in the Job which abended.

[Md.Imtiaz]

I relooked at your question and found that this is the DELETE/DEFINE for VSAM:

Code: Select all

--+----1----+----2----+----3----+----4----+----5----+
DELETE-                                              
 V.PR50.VSAMDT-                                      
  PURGE                                              
   DEFINE CLUSTER(NAME(V.PR50.VSAMDT)CYL(100 20)-    
                 RECORDSIZE(040 1020)-               
        SHR(2)-                                      
        UNIQUE-                                      
                NONINDEXED)-                         
        DATA(NAME(V.PR50.VSAMDT.DATA)CISZ(4096))

That means SHAREOPTIONS is 2.

Is there any other way to know about the VSAM information apart from looking its DEFINE?

[Robert Sample]

If you execute IDCAMS, you can specify LISTCAT to view the definition of a VSAM data set. If you go to option 3.4 of ISPF and use LC on the data name in the list, you can get the same information. The NONINDEXED in the DEFINE means that it is an ESDS and not a KSDS.

The SHAREOPTIONS value gives the cross-region value for jobs running in a single z/OS image, while the second value (if specified -- if not specified, no cross-system access is allowed) gives the cross-system value for jobs running on other z/OS images in the SYSPLEX. SHR(2) which is what you have, with DISP=SHR in the JCL, means you are NOT asking for exclusive access to the data set to write records, so you're pretty much guaranteed to get the problem you see. When writing to a VSAM data set, DISP=OLD in the JCL prevents other jobs / users from accessing the data set while your job is using it. Before you make this change, though, you need to verify what other job(s) are accessing the data set as well since your job cannot begin executing (either the job or that step depending on which JES you use) until you can have exclusive access to it. This may delay your job and potentially impact the overall batch window processing.

SHAREOPTIONS are important when one (or more) jobs can be writing to a VSAM data set while one or more other jobs are reading it at the same time. For example, job A reads key 10000 from a KSDS, updates it, and writes it back to the VSAM data set. At the same time, job B reads key 10000 from the KSDS. Without concurrency control, job B may read the record before job A writes it back to the VSAM and hence retrieved old data. There are multiple ways to deal with this issue; DISP=OLD in the JCL is the easiest. With DISP=OLD, job B won't be allowed to execute until job A is complete (or job A won't be allowed to start until job B is complete; it depends upon which job started first).

[Md.Imtiaz]

Thanks Robert

The job has the folloing message, how would we know it was in contention with which job or CICS program?

Code: Select all

Display 	 Filter 	 View 	 Print 	 Options 	 Search 	 Help                          
 	-------------------------------------------------------------------------------
 	SDSF 	OUTPUT DISPLAY USD025SS JOB32517  DSID   102 LINE  CHARS 'V.PR50.VSAMDT' F
 	COMMAND INPUT ===> 	                                           	SCROLL ===> 	CSR  
    V.PR50.VSAMDT-                                                       000000 
     PURGE                                                               000000 
0IKJ56241I DATA SET V.PR50.VSAMDT NOT ALLOCATED                                 
 IKJ56241I DATA SET IS ALLOCATED TO ANOTHER JOB OR USER                         
 IDC0551I ** ENTRY V.PR50.VSAMDT NOT DELETED                                    
0IDC0001I FUNCTION COMPLETED, HIGHEST CONDITION CODE WAS 8                      
0                                                                               
      DEFINE CLUSTER(NAME(V.PR50.VSAMDT)CYL(100 20)-                     LLV0109
                    RECORDSIZE(040 1020)-                                MCV9701
           SHR(2)-                                                       000000 
           UNIQUE-                                                              
                   NONINDEXED)-                                          LLV0109
           DATA(NAME(V.PR50.VSAMDT.DATA)CISZ(4096))                      000000 
0IGD17101I DATA SET V.PR50.VSAMDT                                               
 NOT DEFINED BECAUSE DUPLICATE NAME EXISTS IN CATALOG                           
 RETURN CODE IS 8 REASON CODE IS 38 IGG0CLEH                                    
 IGD17219I UNABLE TO CONTINUE DEFINE OF DATA SET                                
 V.PR50.VSAMDT                                                                  
0IDC3013I DUPLICATE DATA SET NAME                                               
 	  	DSLIST   	*SDSF    	DSLIST   	-SDSF

[Md.Imtiaz]

If you execute IDCAMS, you can specify LISTCAT to view the definition of a VSAM data set. If you go to option 3.4 of ISPF and use LC on the data name in the list, you can get the same information. The NONINDEXED in the DEFINE means that it is an ESDS and not a KSDS.

Thank you. LC works great.

[Md.Imtiaz]

SHAREOPTIONS are important when one (or more) jobs can be writing to a VSAM data set while one or more other jobs are reading it at the same time. For example, job A reads key 10000 from a KSDS, updates it, and writes it back to the VSAM data set. At the same time, job B reads key 10000 from the KSDS. Without concurrency control, job B may read the record before job A writes it back to the VSAM and hence retrieved old data. There are multiple ways to deal with this issue; DISP=OLD in the JCL is the easiest. With DISP=OLD, job B won't be allowed to execute until job A is complete (or job A won't be allowed to start until job B is complete; it depends upon which job started first).

That means we need to use DISP=OLD in both the jobs?

[Robert Sample]

how would we know it was in contention with which job or CICS program?

If you have console access, you can use command D GRS,C to display contention for data sets. If you don't have console access, you'll need to ask your site support group.

That means we need to use DISP=OLD in both the jobs?

Do both jobs write to the VSAM data set? If so, you should have DISP=OLD on both jobs. If not, the job that writes to the VSAM data set should have DISP=OLD while jobs that just read from it can use DISP=SHR.

[DB2 Guy]

Robert has an excellent explanation, however, why not use the scheduling software to create dependency of jobs on each other?

[Geetu]

I think scheduling software can work BUT if the two jobs are not dependent on each other, apart from using the same file then is it good to DISP=OLD or scheduling software? Will it not confuse in future impact analysis that why these two jobs are dependent of each other?

[Md.Imtiaz]

I think scheduling software can work BUT if the two jobs are not dependent on each other, apart from using the same file then is it good to DISP=OLD or scheduling software? Will it not confuse in future impact analysis that why these two jobs are dependent of each other?

That is why I think what Robert is telling is the best approach.

[Geetu]

We used scheduling changes for the final solution.

[Robert Sample]

I'm glad to hear you got it resolved.

[Geetu]

Thanks for all the help