On Fri, 2 Jul 2010 15:36:53 -0500, Joel C. Ewing wrote: > >When we've had cases where concurrency with DISP=SHR couldn't be avoided >and possible ABEND failure were not acceptable, we have avoided problems >by first staging the PDS data to another dataset and then using a >follow-on job step to copy the data into the PDS with a utility that >uses ISPF SPFEDIT enqueue conventions internally before opening the PDS >for OUTPUT. As long as all batch jobs play by the SPFEDIT enqueue rules >before opening the PDS for OUTPUT, DISP=SHR is reliable. > Why stage rather than write directly to the PDS using ISPF SPFEDIT enqueue conventions.
>If either of the above is not possible, then your best bet may be to >shorten the window of possible interference by adding an initial job >step to your job to copy the needed member(s) to another dataset with an >efficient utility that uses minimal resources, so that your enqueue on >the PDS can be released as quickly as possible before your actual >processing steps begin. > Consider other alternatives, either PDSE or UNIX files, either of which provides a variant of LUW integrity. -- gil ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
