![HP Integrity NonStop NS-series Operation Manual Download Page 210](http://html.mh-extra.com/html/hp/integrity-nonstop-ns-series/integrity-nonstop-ns-series_operation-manual_164441210.webp)
Creating Startup and Shutdown Files
HP Integrity NonStop NS-Series Operations Guide — 529869-005
16 -10
Avoid Manual Intervention
Avoid Manual Intervention
Write startup and shutdown files so that they execute correctly without requiring
manual intervention. Any time an operator must intervene, startup and shutdown time
increase and the possibility of human error increases.
Use Parallel Processing
Parallel processing decreases the time required to start up or shut down your system
or application because startup and shutdown processes are distributed throughout the
processors in your system. For example, this SCF command file uses parallel
processing in four processors to start several communications lines. The files START0,
START1, START2, and START3 contain the actual commands that start the
communications lines.
This command file uses a special technique intended to ensure that each process gets
started even if a given processor is out of service. The technique is to start each
process in two processors. If the first processor is down, the command file continues to
the next processor. If the first processor is up, and the process is started, the
command file still continues to the next processor but fails because the process name
($S
n
) is in use by the process that was successfully started. As a result, a specified
process is started in whichever processor is running. Of course, if neither processor is
up, the attempt to start the process fails.
SCF /IN START0, NOWAIT, CPU 0, NAME $S0/
SCF /IN START0, NOWAIT, CPU 2, NAME $S0/
SCF /IN START1, NOWAIT, CPU 1, NAME $S1/
SCF /IN START1, NOWAIT, CPU 3, NAME $S1/
SCF /IN START2, NOWAIT, CPU 2, NAME $S2/
SCF /IN START2, NOWAIT, CPU 0, NAME $S2/
SCF /IN START3, NOWAIT, CPU 3, NAME $S3/
SCF /IN START3, NOWAIT, CPU 1, NAME $S3/
When using the technique shown in this command file, make sure to spread the
process workload across all available processors. If there are too many processes to
start in processors 0 and 1, queuing and memory-contention problems can result.