I've had interesting situations, when using Allegro's HOURGLAS product.  But all errors were entirely by our own design.

We have a system using a GMT based clock.  Some MPE accounts have an offset private clock in HOURGLAS, based on time zone.  

So for example, our U.S. East Coast has an offset clock of 4 or 5 hours (whether Daylight Savings time or not) so users see their local time.

However when you stream a job when inside the account, it still looks at the system clock for scheduling purposes, and not the account's clock.

Therefore if a JOB has an embedded stream that streams itself in the future, one must ensure that the ;IN= parameter is at least 5 hours or more in the future e.g. ;IN=,5 

Otherwise when the job streams and gets to the point it needs to restream itself again, it decides "Oh, I'm 5 hours late, I need to EXEC right away!"  Within the course of a few minutes, you can have hundreds of instances of the same job launching itself.

Or for example you have a job running at 5PM, and it is set to stream itself or another job at 7PM using the :AT= parameter (;AT=19:00), it really thinks you're at 9PM or 10PM, and it gets put into a SCHED state until Tomorrow. 

Or if the same job is at 2PM, then the job streamed in the future at 7PM is still the same day, so it gets into the SCHED state but goes EXEC in only an hour or right away.

(Above examples are either 4 or 5 hours of difference to GMT based on Savings Time or not.)

Almost enough to give you a headache.

Tracy Johnson
Office (757) 766-4318
[log in to unmask]

* To join/leave the list, search archives, change list settings, *
* etc., please visit http://raven.utc.edu/archives/hp3000-l.html *