> -----Ursprungligt meddelande----- > Från: IBM Mainframe Discussion List > [mailto:[email protected]] För Edward Jaffe > Skickat: den 11 februari 2009 16:41 > Till: [email protected] > Ämne: Re: SV: SV: JCL job execution order > > Thomas Berg wrote: > > BBB should of course have some "anti-parallel" > > mechanism, whether it be the same jobname or if that don't > work, using > > dsn enq's for serializing. > > > > As Bob Rutledge aptly pointed out, duplicate job names is no > guarantee either. > > I do believe that the data set ENQ/DEQ contention mechanism > is strictly FIFO, so it should work to do what you've > suggested. But, keep in mind that BBB will also (presumably) > submit CCC, which will submit DDD and so forth. > > If these jobs run for any significant amount of time at all, > what you'll end up with is every initiator tied up running > jobs that are "stuck" on an ENQ. And when one of these jobs > abends, all of its "successor" jobs will have already been > submitted, will run, and probably fail miserably.
No. 1. BBB will not submit anything until the dsn enq is released. 2. At most there will 2 jobs occupying the initiators, 1 thats executing and 1 thats wating for dataset. (AAA resp BBB, then BBB resp CCC, and so on.) Regards, Thomas Berg __________________________________________ Thomas Berg Specialist IT-U SWEDBANK ---------------------------------------------------------------------- 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
