|
View previous topic :: View next topic
|
| Author |
Message |
jain.eti
New User
Joined: 23 Jan 2007
Posts: 25
Location: delhi
|
|
|
|
Robert, I have not seen any field named record counts in the dump. But I have a field name File #. I believe, it is referring the label on a particular VOL-SER. Is it not so? I initially tried with output dataset with space (100,5), the job got abended with the same message as mentioned above. And then I tried with space for output file as (200,50) but still I am getting the same result.
And Robert, as u said in ur last post that we can find out the space required by record counts, what do u mean by that? Can u give little more explanation so that it will be easier for me to grasp... |
|
| Back to top |
|
 |
Robert Sample
Global Moderator
Joined: 06 Jun 2008
Posts: 8697
Location: Dubuque, Iowa, USA
|
|
|
|
If you divide block size by (average) record length, you get the average number of records per block. Calculate the number of records per cylinder (records per block * 30 if half-track blocking, records per block * 15 if using > half track blocking), divide number of records in the file by records per cylinder to determine number of cylinders required.
Example: 27998 (half track block size) divided by 240 (LRECL from your earlier JCL) = 116 records (truncate anything after the decimal point) minimum per block, times 2 for full track times 15 for tracks per cylinder gives 3480 records per cylinder (minimum -- could be more if average LRECL is less than 240). If you have 500,000 records in the file, that's going to require 144 cylinders (round up) or less, depending on LRECL. At 32760 block size, the same file would require 246 cylinders.
File number would refer to the sequence number of the file on the tape.
Without the message from the system I'm flying blind here, but IF (and that's a big if) you're getting the full 16 extents on the pack you're using 200 + 15 * 50 = 950 cylinders and running out of space. It's also possible you used all free space on the pack and the system couldn't allocate another extent, so you might not have used all that space. If I thought the first case were true, I'd bump the allocation to (1000,200) assuming there's enough space on a pack. if I thought the second case were true, I'd probably try (500,500) to get a smaller primary but more secondary to increase the odds of getting the primary space allocation. I would also probably try to give it another pack or three to expand to should the first selected pack be very short on space. |
|
| Back to top |
|
|
dick scherrer
Moderator Emeritus
Joined: 23 Nov 2006
Posts: 19244
Location: Inside the Matrix
|
|
|
|
Hello,
It will probably help if you post the OUTFILE dd statement that is abending.
How did you get around the s813 abend? |
|
| Back to top |
|
|
enrico-sorichetti
Superior Member
Joined: 14 Mar 2007
Posts: 10873
Location: italy
|
|
|
|
please do not cheat 
now You are talking about a different issue because you
changed the input dataset name
anyway I have always advised against small secondary allocations
allocating lets say (200,20) is a nonsenss
it will simply make the allocation 2.5 times the original, with little room for contingencies
if You have to go for secondary allocations make them as big as the primary!
if You have a tape management system You should be abe to query the record/block count
and with a small effort determine the space needed on disk
or run a repro with a dummy output to get the count |
|
| Back to top |
|
|
jain.eti
New User
Joined: 23 Jan 2007
Posts: 25
Location: delhi
|
|
|
|
Hey Enrico, I am not cheating at all! 
You hurt me! 
Hah! I really did not change the input file name. I am running several jobs per day to complete the work mentioned in my last posts. I have got abend (Sorry not abend but CC = 12) in 3 jobs. The last attached JES message shows those of these Job's JES.
On the other hand, I tried to run the REPRO step for each of these 3 jobs seperately and the message I got earlier was of my seperate run (Seperate run includes only one step to run REPRO). The first JES message attached here shows that of this job's JES message.
So got me now, Enrico?
Yop! Robert showed me how to look into job's description in JES in his 2nd or 3rd post I believe. It was new to me. I analyzed that way and found that orginally these 3 jobs were giving B37. So I attached JES message of those jobs because they refer the problem I am facing not the other one!
SOOOOOOOOOOOOOOOOOOO I AM NOT CHEATING ....    |
|
| Back to top |
|
|
jain.eti
New User
Joined: 23 Jan 2007
Posts: 25
Location: delhi
|
|
|
|
Dick,
I have not got around S813 abend. As I mentioned earlier, whenever I was taking one step of our complete job to check why the complete job ended with CC=12, we received S813. I do not know why! OK! I checked this in CA-1 that the tape is associated with other DSN at label 1. Was it the reason?
But I am confirmed about the label and VOL-SER I used with the particular tape file. And I am confirm that they are pretty fine.
Robert, (Though I am angry wtih you) thanks for your input. I have changed the SPACE with more secondary space. Job is still running and let you all know about its condition once this job gets completed (hopefully this time with CC=0)! |
|
| Back to top |
|
|
jain.eti
New User
Joined: 23 Jan 2007
Posts: 25
Location: delhi
|
|
|
|
Sorry in my last post, it was not Robert but Enrico ...  Sorry for the confusion ... |
|
| Back to top |
|
|
Robert Sample
Global Moderator
Joined: 06 Jun 2008
Posts: 8697
Location: Dubuque, Iowa, USA
|
|
|
|
Well it confused me! 
But if you're reading records and getting SB37 abends then the tape label DSNAME and the JCL DSNAME are matching enough to let the job run. |
|
| Back to top |
|
|
dick scherrer
Moderator Emeritus
Joined: 23 Nov 2006
Posts: 19244
Location: Inside the Matrix
|
|
|
|
Hello,
| Quote: |
| I checked this in CA-1 that the tape is associated with other DSN at label 1. Was it the reason? |
Yes, file #1 would probably be a different dsn. That is not the cause of the s813.
What does CA1 have to say about file #727 on that volume? It has been a while, but i recall being able to list all of the DSNs on a tape using EARL (i prefer that to GRW). Until the mis-match between the jcl and the tape label is resolved, the s813 will continue.
What type of data is on this tape? Having more than 700 user data files on the same tape volume is not something i've seen very often (trying to remember if ever).
If you are still having sb37 abends, it may help to post the OUTFILE dd statement. If we can see what you are using, we can better offer suggestions on how to modify it so it will not abend. |
|
| Back to top |
|
|
enrico-sorichetti
Superior Member
Joined: 14 Mar 2007
Posts: 10873
Location: italy
|
|
|
|
it looks like Your sense of humor is close to 0, haven' t You seen the smiley
I just went to see my ophthalmologist and he confirmed that my sight is good
can You say the same ?
in the first post you have
| Quote: |
| SYS3.QPP1D.ARCH.D08118.T112723.F01 |
and in the second
| Quote: |
| SYS3.QPP1D.ARCH.D08119.T105510.F01 |
so You might have been using the same datasets, but You posted something different
being angry is just a personal choice, nobody forced You to |
|
| Back to top |
|
|
gcicchet
Senior Member
Joined: 28 Jul 2006
Posts: 1702
Location: Australia
|
|
|
|
Hi,
I don't know if it's just me, but I'm confused with this whole thread, so let's see if I can add more to it.
Firstly, why is the file written to disk as a temporary dataset, what is going to read the temporary dataset ?
Why not process directly from tape ?
Blocksize of 32760 is very inefficient, 27998 will improve space usage by about 40%
Gerry |
|
| Back to top |
|
|
dick scherrer
Moderator Emeritus
Joined: 23 Nov 2006
Posts: 19244
Location: Inside the Matrix
|
|
|
|
Hi Gerry,
Welcome aboard. . .
Been so long since i've looked at the front of the topic, i completely missed/forgot that the OUTFILE dd was there. . .
Jain, Sorry 'bout that (the dd info request).
As you will probably continue to write to dasd for ftp (i suppose it might run from tape, but i've never tried it), suggest you change this:
| Code: |
// UNIT=TEMPDA,
// SPACE=(CYL,(100,5),RLSE),
// RECFM=VBA,LRECL=240,BLKSIZE=32760
|
to
| Code: |
// UNIT=(TEMPDA,4),
// SPACE=(CYL,(100,100),RLSE),
// RECFM=VBA,LRECL=240
|
You might also consider using the sort product for the copy as it will most likely run faster than IDCAMS. |
|
| Back to top |
|
|
|
|
