Based on the things you have said you cannot do, I think the only solution so far is to have all steps in the same job.
David Logan Manager of Product Development, Pitney Bowes Business Insight http://centrus.com W: (720) 564-3056 C: (303) 818-8222 -----Original Message----- From: IBM Mainframe Discussion List [mailto:[email protected]] On Behalf Of Dave Salt Sent: Wednesday, February 11, 2009 07:40 To: [email protected] Subject: Re: JCL job execution order > From: [email protected] > That's my reading as well: Job AAA must complete (successfully or not) > before Job BBB begins execution; Job BBB must complete (successfully or > not) before Job CCC begins execution, and so on. I.e., it doesn't > matter whether none, some or all of the jobs complete successfully, so > long as all jobs execute sequentially in the order submitted. The above is correct; i.e. AAA must complete (successfully or not) before BBB begins execution, and so on. My hope was to do this without modifying the JCL that's currently being generated, or at least keeping the modification to a minimum. The reason is the user might still want to manually submit individual jobs from time to time, so I don't want to change the JCL if I can help it. For example, I'm not overly enthusiastic about adding a step to the end of each job that submits the next job. Because this automation is for an end user, there is no chance of changing the way JES works (e.g. adding a new job class or whatever). I had thought about adding a wait time between submitting each job (e.g. 10 seconds), but I'm still not sure how well this would guarantee the jobs execute in the correct order. The same thing applies to using DISP=OLD on a data set; yes it would guarantee that only 1 job can execute at a time, but it still wouldn't guarantee that the jobs execute in the right order. I'm starting to think that maybe I can allow the existing process to generate the JCL, but my process would pass it a switch that tells it not to submit the job. A fraction of a second later my process would call the existing process to generate the next job, and the switch would tell the existing process to add the next job (with no job card) to the bottom of the previous job (and so on). When my process calls the existing process to generate the last job, it would set the switch to tell the existing process to submit the generated job. The only messy part might be ensuring the step names are unique throughout the job, but other than that I think this might be the best solution? My thanks to everyone who offered their suggestions on this! Dave Salt SimpList(tm) - try it; you'll get it! http://www.mackinney.com/products/SIM/simplist.htm _________________________________________________________________ So many new options, so little time. Windows Live Messenger. http://www.microsoft.com/windows/windowslive/products/messenger.aspx ---------------------------------------------------------------------- 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 ---------------------------------------------------------------------- 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
