Unicenter CA-Scheduler Job Management for VSE User Guide
Unicenter CA-Scheduler Job Management for VSE User Guide
Unicenter CA-Scheduler Job Management for VSE User Guide
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
7.5 Multiple CPU Considerations<br />
tracked there. Do not <strong>for</strong>get, the default SYSID is the first one<br />
in your SYSID list specified when generating <strong>Unicenter</strong><br />
<strong>CA</strong>-<strong>Scheduler</strong> (<strong>CA</strong>IJGEN macro) and will be used <strong>for</strong> any<br />
schedule defined without a SYSID.<br />
If the job is actually executed on a CPU other than the<br />
controlling one the other CPU will notify the controlling CPU<br />
of any in<strong>for</strong>mation it would have picked up itself, such as job<br />
started and job completed, along with the termination code.<br />
This is done by writing in<strong>for</strong>mation on the tracking file to be<br />
passed along as internal events. If any job is defined without<br />
a SYSID, the default SYSID is the one specified (or defaulted)<br />
at the schedule definition.<br />
Operator Commands<br />
When you issue operator commands, there are times that you<br />
should be aware of which CPU is in control of the schedule or<br />
job. That is, if you issue a <strong>CA</strong>NCEL operator command <strong>for</strong> a<br />
job that is not controlled on the CPU on which you are on, it<br />
is written to the tracking file and within seconds (could be up<br />
to 30 seconds), processed by the CPU controlling that job. If,<br />
<strong>for</strong> some reason, the CPU is not able to handle it right away<br />
(suppose it is down), as soon as it can, it will process the<br />
command; even if it is hours later. The area on the tracking<br />
file that passes these operator commands and status<br />
in<strong>for</strong>mation back and <strong>for</strong>th between CPUs is called the<br />
Inter-Communication Records area (ICR).<br />
The only operator command that cannot work in a multi-CPU<br />
environment is the STATUS command using the priority<br />
sequence. Since the priority order happens to be an in-core<br />
mechanism, <strong>for</strong> per<strong>for</strong>mance reasons, each CPU has its own<br />
list. Thus, when you request a display of jobs on another<br />
CPU, the priority option cannot be used.<br />
Do keep in mind that any operator commands that are issued<br />
that have to execute on a CPU other than the one they were<br />
issued from, will be sent across the ICR on the tracking file.<br />
They will remain there until actually processed or purged.<br />
When the down CPU becomes operational again and<br />
<strong>Unicenter</strong> <strong>CA</strong>-<strong>Scheduler</strong> is subsequently started, <strong>Unicenter</strong><br />
<strong>CA</strong>-<strong>Scheduler</strong> determines if it missed an autoscan. If it missed<br />
an autoscan, <strong>Unicenter</strong> <strong>CA</strong>-<strong>Scheduler</strong> immediately per<strong>for</strong>ms<br />
an autoscan. If, during this autoscan, <strong>Unicenter</strong> <strong>CA</strong>-<strong>Scheduler</strong><br />
finds ICRs that are more than 24 hours old, it purges those<br />
old ICRs without executing them.<br />
Another consideration <strong>for</strong> multiple CPU environments is the<br />
method <strong>for</strong> executing jobs at a CPU other than the controlling<br />
CPU. To do this on a permanent basis, specify the SYSID of<br />
the CPU at which the job will execute in the RUN ON SYSID<br />
field of the job's base record. To alter the CPU <strong>for</strong> one<br />
7-22 <strong>Unicenter</strong> <strong>CA</strong>-<strong>Scheduler</strong> <strong>User</strong> <strong>Guide</strong>