OpenVMS Job Scheduler - Recomendations

[B Claremont]

Sorry for the incomplete previous post. It occured to me that virus scanners might not like an attachment with a .COM extension, so I combined both sample SUBMIT procedures into the attached text file. SUBMIT.COM provides an example of the DCL SUBMIT command passing a few parameters. AFTER_HOURS_BATCH_SETUP.COM is an example of a batch procedure that resubmits itself.

[Jan van den Ende]

John,

we essentially follow Steven's idea:
self re-submiiting batches.
To steer all kinds of scheduling, we wrote a DCL routine (made available at DCL.OPENVMS.ORG):

http://dcl.openvms.org/stories.php?story=04/10/15/8590853

@ with the scheduling wishes.

What can be specified, and the correct syntax for specifications, is documented in the header of the routine.


hth

Proost.

Have one on me.

jpe

[Thomas Ritter]

One of the main reasons we like the CA Scheduler is the GUI. It makes light work for Operations staff to place jobs on hold, change scheduled times, change dependancies, release jobs. We have a number of environments within the VMS cluster and during some environment outages some job streams have to be suspended. It is not the technical VMS staff which run the batch processing environments, but the Operators. This should be your focus. Who will ultimately have to manage the work flow ?

AUS 10 cents.

[Wim Van den Wyngaert]

We use JAMS from ARGENT.

Whatever you use be careful about side efects.

1) how does the system react when you abort the job ? (jams abort the schedule of e.g. every 15 minutes too).

2) how does the system react when you manually submit the job file ? (jams starts the schedule, even when it was already active).

3) can you prevent 2 jobs with the same name (and work files) from running simultaniously ?

4) Can the scheduler handle job names for which the logical name is not cluster wide defined and how does it react in cluster failover ?

5) How do you know that a job was NOT started ?

6) How does it react on time changes ?

7) how does it react after a cluster down time of a few hours ? (jams starts all jobs at the same time).

Wim

[William Webb_2]

I highly recommend JAMS.

We had over 20,000 batch jobs running per night at one of my former sites, and it ran like gangbusters.

We also did many system consolidations, and the ability to export a system's entire JAMS job and setup information, and then import it onto the target system saved us hundreds if not thousands of man-hours that keying all that stuff in would have involved.

Their tech support's great, too.

Wim is right about things you have to look out for, but most of those are going to be gotchas no matter which product you use.

WWWebb

[Thomas Ritter]

John, some really good information has been provided. Would you care to assign points ? If you feel a response is worthless assign zero points.

Years ago when we made our decision to buy a scheduler these forums were not available.

[John T. Farmer]

Ok, this may sound a little strange, but does anyone have a port of CRON/KRONOS for VMS done in COBOL? That is currently the only language we have on our Alpha with VMS7.2. I want to modify the Cron code a bit and have no access to C or Fortran (forgot which it is written in). Thanks for any ideas on this subject.

John
john dot farmer at genworth dot com

[Karl Rohwedder]

Perhaps you may use one of the testdrive systems (http://www.testdrive.hp.com) to modify the source to your needs. The CRON utility I packaged is written in C (http://www.the-rohwedders.de/downloads/index.html).

regards Kalle

[Dean McGorrill]

Hi John,
with so small amount of stuff on your
system, it might be a good time to upgrade
before you put too much on it (I'm stuck
at 7.2-1) for simple schduling I've always
just dcl'ed it. If I had 20000 jobs I wouldn't be using dcl! good luck -Dean