View previous topic :: View next topic
|
Author |
Message |
spoorni
New User
Joined: 29 Nov 2007 Posts: 20 Location: india
|
|
|
|
Recently i had a job that failed in STEP04. There were 3 output files from STEP04 - 10,20,752. The DISP parameters for 20 and 752 was (NEW,CATLG,DELETE). The DISP parameter for 10 was (NEW,PASS,DELETE). Once the job abended with a user abend, the files for 20 and 752 were still existing. But the file for 10 was deleted. As per my understanding, the last DISP parameter is for abnormal termination. Depending on that, the files for 20 and 752 also should have been deleted. Is that correct? Also, do DISP parameters work only for system abends and not user abends? |
|
Back to top |
|
|
Akatsukami
Global Moderator
Joined: 03 Oct 2009 Posts: 1788 Location: Bloomington, IL
|
|
|
|
Your account of the JCL is incoherent. Please post it, together with the messages for the step from the JES message log, enclosed in Code tags. |
|
Back to top |
|
|
steve-myers
Active Member
Joined: 30 Nov 2013 Posts: 917 Location: The Universe
|
|
|
|
- In z/OS a “file” is the data on magnetic tape between the beginning of the tape and a “file mark”, or the data between two “file marks.” Period. End of story. There are no “files” on disk storage. z/OS uses “data sets.” A “file” in toy and baby systems is an unstructured stream of bytes, compared to a “data set,” which has a structure.
- When you specify DISP=(NEW,PASS), the data set will usually be deleted at the end of the job unless a DISP parameter for the same data set in a later step that is executed specifies DISP=(OLD,KEEP) or DISP=(OLD,CATLG).
- The conditional DISP only applies to the step being executed. It will not apply if a different step abnormally terminates.
- The conditional DISP applies only if the step terminates with an ABEND. Terminating with a return code that is not 0 is not an ABEND. If the program has some mechanism to recover from an ABEND, the step does not terminate with an ABEND, so the conditional DISP does not apply.
- It does not happen very often, but if your program damages the address space to the point where the address space fails, conditional DISP is not done.
|
|
Back to top |
|
|
nevilh
Active User
Joined: 01 Sep 2006 Posts: 262
|
|
|
|
Quote: |
There are no “files” on disk storage. z/OS uses “data sets.” |
I would tend to disagree with this. In Europe at least it is quite normal to speak of input files and output files. It makes no difference if they are on tape or disk. The IBM literature makes reference to Disk files so I am happy to go along with this. |
|
Back to top |
|
|
steve-myers
Active Member
Joined: 30 Nov 2013 Posts: 917 Location: The Universe
|
|
|
|
nevilh wrote: |
Quote: |
There are no “files” on disk storage. z/OS uses “data sets.” |
I would tend to disagree with this. In Europe at least it is quite normal to speak of input files and output files. It makes no difference if they are on tape or disk. The IBM literature makes reference to Disk files so I am happy to go along with this. |
True, but the use of the terminolgy is incorrect. When you say "IBM literature," please be more precise. I also recall the use of "file" in the formal documentation of control units such as the 2841, but the context usually refers to the disk drive as a whole, not to the data on the disk as recognized by DOS/360 or OS/360.
I will tolerate the use of "disk files" in the context of a "file" in the pseudo Unix environment, where you have the usual disorganized mess of a Unix "file."
IBM went out of its way to avoid a disorganized mess. They fell into this trap with the 14xx and the more character oriented 7000 series computers, and appear to have made a deliberate decision to avoid it with the System/360 mainline operating systems.
I do believe they made a mistake when they pushed fixed length records as more of a default rather than the VB data set format, though it did have the desirable effect (at least from IBM's point of view) selling a lot more disk drives and tape media.
I also believe count-key-data disk drive architecture, at least as implemented in the mainline operating systems, was a goof of historic proportions, though that is a topic that should be discussed elsewhere. I do know when they started in on fixed block architecture disk drives in the 1980s I was hoping they would push it into MVS and perhaps VSE. The rise of RAID and disk caching in the control units ultimately hid the serious performance issues with count-key-data architecture. |
|
Back to top |
|
|
nevilh
Active User
Joined: 01 Sep 2006 Posts: 262
|
|
|
|
Authorised Assembler Services Guide say's
Quote: |
The operator can start utility programs to read jobs from disk or tape
files and submit them to the system. |
and the DFSMS courses nowadays all refer to VSAM Files . |
|
Back to top |
|
|
Ed Goodman
Active Member
Joined: 08 Jun 2011 Posts: 556 Location: USA
|
|
|
|
To answer the TS's last question, the 'bad result' DISP parm applies to abends, not just bad return codes.
The file for your '10' DD didn't stick around because it was never cataloged in the first place. |
|
Back to top |
|
|
steve-myers
Active Member
Joined: 30 Nov 2013 Posts: 917 Location: The Universe
|
|
|
|
Ed Goodman wrote: |
... The file for your '10' DD didn't stick around because it was never cataloged in the first place. |
Ed, that's hard to know.
Before SMS,
//A EXEC PGM=...
//xx DD DISP=(NEW,PASS),...,DSN=xx
//B EXEC PGM=...
//xx DD DISP=(OLD,CATLG),DSN=xx
The data set was not cataloged until step B completed.
With SMS, the data set is cataloged when it is allocated. It will be uncatalogued and deleted if step B does not run. |
|
Back to top |
|
|
Ed Goodman
Active Member
Joined: 08 Jun 2011 Posts: 556 Location: USA
|
|
|
|
Good point. |
|
Back to top |
|
|
|