The support for an auto-scheduled, unsupervised "job" on Linux/Unix and Windows seems much more limited than on MVS, where multiple semi-independent steps and possible restart after failure at various steps within a job, and automated scheduling of dependent jobs are the norm. The sometimes-frustratingly-limited MVS JCL, which used to make all the data flow in and out of each job step obvious before the days of databases, also contributed to the ability to build automated job schedulers and job restart managers that could provide support for job restart at many internal job steps in the event of a job failure.
In Linux and presumably also in Windows. a scheduled "job" process that fails must generally be manually re-run in its entirety and designed to support that. You could design a complex script with many internal steps in a way that it might "detect" which steps had successfully completed and not repeat them on a re-run, but the logic to do that would have to be uniquely designed and written for each script. The mainframe restart approach was necessary in order to support production jobs or job sequences that ran for hours in environments where restart from the beginning would exceed the available time or simply be too costly in resources. Typically, scheduled processes on the other platforms did not run long enough to demand partial re-run support. To make automated scheduling on any platform practical for time-critical production work, there needs to be someone monitoring the system closely enough to detect failures and see that they are resolved, and that the production process completes by a specified deadline. In reality, auto-scheduled critical jobs on an MVS system are always loosely "supervised" by an operations staff that monitors 100's or even 1000's of daily jobs and on-line systems for failures, and who invoke the services of any other personnel needed to resolve a failure. That MVS oversight can be leveraged to have scheduled jobs/processes on MVS initiate tasks on remote systems and have failures of those remote tasks induce failure alerts on MVS that will be acted on by the same operations staff. With that approach, any automation available on MVS can be used to schedule and monitor tasks on a remote server with conditions and dependencies more sophisticated than the native scheduling support on those other platforms. Joel C Ewing On 12/22/20 5:57 PM, John McKown wrote: > On Tue, Dec 22, 2020 at 5:49 PM Seymour J Metz <[email protected]> wrote: > >> https://en.wikipedia.org/wiki/Cron >> >> > I use CRON a fair amount on Linux at home. And on z/OS at work, for > "personal scheduling". And there's the equivalent "Windows Task Scheduler" > on Windows. But I wonder if any business uses them for things that we use > CA-7 for. IIRC, CA had a Windows product where you'd put an "agent" on a > Windows system and CA-7 could schedule work there. We actually tried to use > it. But it never caught on in the Windows world at my company. > > There is this: > https://docs.microsoft.com/en-us/sql/reporting-services/install-windows/configure-the-unattended-execution-account-ssrs-configuration-manager?view=sql-server-ver15 > > But I'm getting too far off topic. > > > >> -- >> Shmuel (Seymour J.) Metz >> http://mason.gmu.edu/~smetz3 >> >> ________________________________________ >> From: IBM Mainframe Discussion List [[email protected]] on behalf >> of John McKown [[email protected]] >> Sent: Tuesday, December 22, 2020 6:35 PM >> To: [email protected] >> Subject: Re: JCL divergence >> >> On Tue, Dec 22, 2020 at 5:26 PM Gibney, Dave <[email protected]> wrote: >> >>> Windows....Job...Huh >>> >> Right. I am so ignorant, perhaps the Windows (and Linux?) world doesn't >> even have any unattended scheduled activities. I know that there is a >> "Windows Scheduler" that can run a batch file (MSDOS .bat) automatically at >> a given time or when a particular user does a "log on" (perhaps akin to a >> TSO logon proc). >> >> If the above is true (no automated unattended work), I wonder how companies >> do {month,quarter,year}-end processing to generate reports to send to >> appropriate governmental bodies. Or even, as in my employer, to >> policyholders. >> >> >> >>>> Hum, how do the Windows experts "restart" a "job" that fails? I really >>>> don't now. >>>> >>> ... -- Joel C. Ewing ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
