On Tue, Jan 31, 2012 at 7:52 PM, Scott Ford <[email protected]> wrote: > Sam, > > That's a interesting idea...
Scott You also need to handle the case like: 1) The submitted job gets a JCL error prior to executing your code; STC is never posted. 2) The submitted job gets queued up on resource contention; STC is left waiting for a long time. 3) The submitted job is forced off the system in a way that precludes the ESTAE from being invoked; STC is never posted. I think these can be solved by attaching a sub-task in the STC to monitor/wait for the submitted job to complete. The sub-task would need a stimer to handle the case where the submitted job does not post back. After a certain period of time, it is assumed that the submitted job failed, and then recovery processing or corrective actions are taken. Sam > > Sent from my iPad > Scott Ford > Senior Systems Engineer > www.identityforge.com > > > > On Jan 31, 2012, at 10:32 PM, Sam Siegel <[email protected]> wrote: > >> Scott - Does the submitted job run authorized? If it does, you can >> use a combination of name token service and pause element service to >> coordinate tcb in different address spaces. >> >> You would need to code an estae or have a resmgr active to handle >> error situations and ensure that the STC is not left waiting. >> >> The submitted job would have to run on the same mvs image as the STC. >> >> Sam >> >> On Tue, Jan 31, 2012 at 6:37 PM, Scott Ford <[email protected]> wrote: >>> Scheduler is out of the question, we are self-contained within a STC. >>> >>> >>> Sent from my iPad >>> Scott Ford >>> Senior Systems Engineer >>> www.identityforge.com >>> >>> >>> >>> On Jan 31, 2012, at 9:18 PM, Lizette Koehler <[email protected]> >>> wrote: >>> >>>>> >>>>> A timed approach is not great, how long do you wait, the submitted job >>>> creates a file... >>>>> >>>>> Sent from my iPad >>>>> Scott Ford >>>>> Senior Systems Engineer >>>>> www.identityforge.com >>>>> >>>> >>>> So the question becomes what are you trying to accomplish. >>>> >>>> Many scheduling packages have file triggers and dataset triggers that >>>> monitor the system (using SMF records) to "see" a dataset that is created. >>>> >>>> So are you trying t invent this type of process? If so, you should provide >>>> us with more details on what you are doing. >>>> >>>> A job that has a last step that does something may be a simpler process. >>>> >>>> Lizette >>>> >>>> ---------------------------------------------------------------------- >>>> For IBM-MAIN subscribe / signoff / archive access instructions, >>>> send email to [email protected] with the message: INFO IBM-MAIN >>> >>> ---------------------------------------------------------------------- >>> For IBM-MAIN subscribe / signoff / archive access instructions, >>> send email to [email protected] with the message: INFO IBM-MAIN >> >> ---------------------------------------------------------------------- >> For IBM-MAIN subscribe / signoff / archive access instructions, >> send email to [email protected] with the message: INFO IBM-MAIN > > ---------------------------------------------------------------------- > For IBM-MAIN subscribe / signoff / archive access instructions, > send email to [email protected] with the message: INFO IBM-MAIN ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
