View previous topic :: View next topic
|
Author |
Message |
Selva-kumar
New User
Joined: 01 Mar 2007 Posts: 52 Location: chennai
|
|
|
|
Hi,
I'm trying to do a repro on a KSDS file to a flat file (VB). The job abend with EOV abend and only few records ae copied. When i restarted again, the job ran succussfully.
Here are the attributes of the files:
Input KSDS file:
Code: |
KEYLEN---------------150 AVGLRECL-------------250
RKP--------------------0 MAXLRECL-------------900
STATISTICS
REC-TOTAL--------5533469 SPLITS-CI---------126033
REC-DELETED----------742 SPLITS-CA-----------2247
REC-INSERTED-----1096352 FREESPACE-%CI----------0
REC-UPDATED------1744924 FREESPACE-%CA----------0
REC-RETRIEVED-2147483647 FREESPC-------1249091584
USER-DATA-SIZE-----------------------------3252214895
ALLOCATION
SPACE-TYPE------CYLINDER HI-A-RBA------3204218880
SPACE-PRI------------888 HI-U-RBA------2952069120
SPACE-SEC------------494
|
Attributes of Output flat file:
Code: |
UNIT=SYSDA
DCB=(RECFM=VB,LRECL=0904,BLKSIZE=27998)
SPACE=(CYL,(0100,0100),RLSE
|
When i analyzed the two output files (one during abend and one during successful execution), it shows as "Number of volumes allocated as 5".
But in the first case (abend), 5 volumes are allocated and all are 100% utilized.
In the second case (success), it uses only 3 volumes and 99% is utilized.
I like to know that what could be the case in which the job ran successfully in the second time.
Info on both the datasets are given below. dataset1 - created on abend;dataset2 - created on successful completion
Code: |
Command - Enter "/" to select action Tracks %Used XT
-------------------------------------------------------------------------------
Dataset1 55560 100 43
Dataset2 59160 99 40
***************************** End of Data Set list ****************************
|
|
|
Back to top |
|
|
expat
Global Moderator
Joined: 14 Mar 2007 Posts: 8797 Location: Welsh Wales
|
|
|
|
It would appear that volume fragmentation could be the cause of your abend here |
|
Back to top |
|
|
PeterHolland
Global Moderator
Joined: 27 Oct 2009 Posts: 2481 Location: Netherlands, Amstelveen
|
|
|
|
My guess is that your blocksize is not a multiple of 904 plus 4. |
|
Back to top |
|
|
Selva-kumar
New User
Joined: 01 Mar 2007 Posts: 52 Location: chennai
|
|
|
|
Expat,
What could be done to prevent this. can the allocation of the file be increased? |
|
Back to top |
|
|
expat
Global Moderator
Joined: 14 Mar 2007 Posts: 8797 Location: Welsh Wales
|
|
|
|
Peter, I doesn't have to be for a VB file.
I always use 27998 for VB files |
|
Back to top |
|
|
Selva-kumar
New User
Joined: 01 Mar 2007 Posts: 52 Location: chennai
|
|
|
|
Hi,
I also like to add up the following point. In first case (when job abends), the extents in 5 volumes are as follows:
Vol1 - 4 extents
Vol2 - 1 extent
Vol3 - 6 extent
Vol4 - 16 extent
Vol5 - 16 extent
In the second case(when job is successful), the extents in the 3 volumes ae as follows:
Vol A - 16 extent
Vol B - 16 extent
Vol C - 8 extent
These three volumes are different from the first 5 volumes that were used. |
|
Back to top |
|
|
expat
Global Moderator
Joined: 14 Mar 2007 Posts: 8797 Location: Welsh Wales
|
|
|
|
expat wrote: |
It would appear that volume fragmentation could be the cause of your abend here |
|
|
Back to top |
|
|
PeterHolland
Global Moderator
Joined: 27 Oct 2009 Posts: 2481 Location: Netherlands, Amstelveen
|
|
|
|
Expat,
thats right. After scanning some of my jcl libraries i found the same value for VB. Probably a mental remainder from the time i was an ICT virgin. |
|
Back to top |
|
|
Selva-kumar
New User
Joined: 01 Mar 2007 Posts: 52 Location: chennai
|
|
|
|
Thanks expat. I like to know how this can be prevented in future runs. |
|
Back to top |
|
|
PeterHolland
Global Moderator
Joined: 27 Oct 2009 Posts: 2481 Location: Netherlands, Amstelveen
|
|
|
|
Selva,
you could use dedicated volumes for your output. Or ask the storage people to do something about fragmentation. |
|
Back to top |
|
|
Selva-kumar
New User
Joined: 01 Mar 2007 Posts: 52 Location: chennai
|
|
|
|
I think i need to ask the storage group folks to do something to take care of the fragmentation. Thanks Peter
It would be helpful if you could point me any materials to understand the storage and volume allocation. |
|
Back to top |
|
|
expat
Global Moderator
Joined: 14 Mar 2007 Posts: 8797 Location: Welsh Wales
|
|
|
|
If your storage people were anywhere near decent this fragmentation should not occur |
|
Back to top |
|
|
Pete Wilson
Active Member
Joined: 31 Dec 2009 Posts: 582 Location: London
|
|
|
|
You could increase the volume count by specifying VOLUMES(* * * * * * * * * *) in your define (same as VOL=(,,,10) in JCL) so the dataset can extend over further volumes as required. Sometimes doing that and reducing your allocation size a bit can help when you have fragmented volumes.
It might also help to assign a Dataclas that gives Extended Format as that increases the 16 extents per volume limit to 123 and removes the 65k tracks per volume size limit as well.
Remember that the old rules about the primary or first secondary extend having to be achieved in 5 chunks doesn't apply under later versions of SMS, they could be taken in many extents because of Space Constraint Relief. Add to that the automatic extent consolidation where two extents may be 'joined' on a subsequent extend if there's contiguous space next to the previous one and it starts to get a bit difficult to make sense of extents sometimes now. |
|
Back to top |
|
|
|