CODEWITHSUNDEEP
mainframedfsort
Parushuram
Parushuram

Reputation: 41

can you please explain me the difference between BLKSET sort option and NOBLKSET sort option?

Recently I came across an abend in a SORT step in a Mainframe job, where SORTOUT is VSAM file and SORTIN is a equential file.

The error is: 

ICE077A 0 VSAM OUTPUT ERROR L(12) SORTOUT

One of my senior colleague suggested to me to see if there are any duplicates, but I didn't find any duplicates in the input file. s After going thru some manuals, I found that OPTION NOBLKSET control card overrides the default BLOCKSET COPY TECHNIQUE, and can be used to bypass sort errors (provided all the possible effects of bypassing the sort error are analysed), so I used OPTION NOBLKSET. Now the step executes successfully.

After analysing the SYSOUT I found that 

ICE143I K PEERAGE      SORT  TECHNIQUE SELECTED

Can any one explain how the BLOCKSET technique works and how PEERAGE technique works? SORT used in our system is DFSORT.

Upvotes: 0

Views: 1984

Answers (1)

Bill Woodger
Bill Woodger

Reputation: 13076

You can start here, which explains that of three techniques Blockset is DFSORT's preferred and most efficient technique for sorting, merging and copying datasets: http://pic.dhe.ibm.com/infocenter/zos/v1r12/index.jsp?topic=%2Fcom.ibm.zos.r12.icea100%2Fice1ca5028.htm

Peerage/Vale and Conventional are the other two techniques, of which one is selected which is thought to be the next best if it is not possible to use Blockset.

You have misread references to the use of NOBLKSET. In cases where what would effectively be "internal" errors encountered by DFSORT, and if BLOCKSET is being used, turning off Blockset will cause another SORT method to be chosen, which perhaps will get your step run and Production finished whilst the error is investigated with the step that used Blockset.

NOBLKSET is not a cure-all and did not affect your use of DFSORT. You should only use NOBLKSET in very limited circumstances which are suggested to you for very particular reasons. Blockset is significantly more efficient than Peerage/Vale or Conventional.

You should update your question with a sample of the input data and an IDCAMS LISTCAT of the KSDS.

You either had a duplicate key, or the insertions (your file being written) were not in sequence. Remember you can get duplicates if you have a KSDS with data on it already.

If you want details about Blockset and Peerage/Value, you'll have to hit technical journals and possibly patent listings. I don't know why you'd want to go that far. Perhaps knowing that, you now don't?

Upvotes: 2

Related Questions