Safeguard Reference Manual (G06.24+, H06.03+ )

Event-Exit-Process Commands
Safeguard Reference Manual—520618-013
15-34
Event-Exit Design, Management, and Operation
If the event-exit process stops abnormally, $ZSMP attempts to restart the process until
it is successfully restarted or disabled. An EMS message is sent to the console
informing the operator each time a restart is attempted. $ZSMP attempts each restart
at an interval designed to avoid a negative impact to the performance of the $ZSMP.
Messages are not be sent to the event-exit process until the restart is successful.
If the processor in which the event-exit process is running becomes unavailable, the
backup CPU is used as the primary CPU. If neither the primary or backup CPU is
available, the event-exit process is restarted in the same CPU as the $ZSMP.
$ZSMP opens the event-exit process with the name $name.#ZSEEP, where name is
the name assigned in the event-exit configuration record. Because the event-exit
process can be opened by processes other than Safeguard security processes, the
subdevice name #ZSEEP allows the event-exit process to determine the intention of
the opener. It also allows the event-exit process to determine which message protocol
is being used.
To avoid deadlocks, the event-exit process is a security process. Messages received
from the event-exit process are not be sent to the event-exit process. Child processes
of the event-exit process are not security processes. To avoid deadlocks, the event-exit
process must maintain an internal list of its child processes and not forward their own
requests to them.
Once the event-exit process has responded to the open request of an SMON, it must
not perform waited I/O.