How to avoid SB37 abend in ICETOOL?

Ramanan-R · Posted: Thu Oct 25, 2007 8:52 pm

Hi All,

I tried all the suggestions given in this forum, but none is working for me.

Frank Yaeger · Posted: Thu Oct 25, 2007 9:16 pm

You haven't given any information that identifies the data set which got the B37 or what you're doing.

Show your JCL and your JES, //TOOLMSG and //DFSMSG output.

Ramanan-R · Posted: Thu Oct 25, 2007 9:31 pm

Frank,

Please find the details below,

JCL:

Frank Yaeger · Posted: Thu Oct 25, 2007 10:12 pm

You didn't show the JES messages that identify the data set that got the B37, but it appears from the other messages that it's the T1 data set.

It appears to be a case of just not having enough space for the T1 output records. It may be that the volumes selected just don't have the needed space available. I'm not an expert on this kind of thing, so maybe somebody else has some ideas.

For referenace, approx. how many records does IN1 have? IN2?

I did notice a few things though:

You seem to only need 600 bytes of the 1200 byte records. So you could cut down the amount of space needed by changing your INREC statements to:

Ramanan-R · Posted: Fri Oct 26, 2007 11:08 am

Frank,

Thanks for your suggestions.

I tried by making the changes you mentioned, but problem still persists.

There are around 4 million records in both the input files.

JCL:

gcicchet · Posted: Fri Oct 26, 2007 11:18 am

As Frank asked earlier, why can't you provide the entire job output ? JES messages etc

Gerry

Ramanan-R · Posted: Fri Oct 26, 2007 4:01 pm

Please find the entire JES output below,

Frank Yaeger · Posted: Fri Oct 26, 2007 11:11 pm

It managed to copy the IN1 records to T1, but got the B37 trying to copy the IN2 records to T1.

The following seem to be space related messages for T1, but I don't know what product is issuing them or what they mean:

dick scherrer · Posted: Sat Oct 27, 2007 12:22 am

Hello,

You could talk with the people who are responsible for storage management for your system. There may not be enough space on UNIT=DISK. The storage management people can tell you if there is another UNIT= you can use for this requirement. You will need to tell them the size and number of records you are working and the expected size of the final output.

It may also be that you need to split the original input files, process the smaller sets, and then merge them back together later.

gcicchet · Posted: Mon Oct 29, 2007 8:23 am

Hi,
looking at the output you have posted it appears that the T1 DDNAME is not using the additional volumes you have specified.

One way to prove this is to create a permanent dataset instead of a temporary one and changing the disp to MOD,CATLG,CATLG.

After the job completes (abends), under ISPF 3.4 you should be able to see the number of volumes used.

If the dataset is not using more than 1 volume, this is your problem assuming you are using 3390's as disk space.

The total number of cyls required for 8 million records should be about 6000 cyls (almost 2 3390's )
You can always write the output to cart.

Gerry

reeno_sj · New User Joined: 21 Mar 2007 Posts: 1 Location: chennai

Hi,

I faced the similar problem and changing the UNIT=(WORK,8) solved this problem. My input records were also having 9.3 million recordsand I used UNIT = SYSDA previously. Hope this might solve your problem.