NET/MASTER Management Services (MS) System Management Guide

Managing NCL Queues
Customizing the Operations Environment
115414 NonStop NET/MASTER MS System Management Guide 5–21
Overview of NCL Queues NCL processes can make logical decisions based on the analysis of system and
network information from event messages. This feature of NCL, coupled with the
ability of an NCL process to monitor and control system and network activity and to
react automatically to events, allows you to automate many NonStop NET/MASTER
MS operations. For example, an NCL process can automatically perform tasks such as
reloading a failed CPU, rebalancing CPUs after reloading, restarting devices and
subdevices, restarting failed Expand and SNAX/XF lines, restarting failed Pathway
terminals, restarting stopped TACL processes, and so on.
However, if numerous failures occur concurrently or frequently, many NCL processes
can start that try to recover from the failures: one NCL process for each failure. If the
number of concurrently executing NCL processes continues to increase, NonStop
NET/MASTER MS may experience severe performance degradation due to increased
memory consumption.
You can use NCL queues to control the number of concurrently executing NCL
processes executed by the START verb from a central controlling NCL process, such as
an EMSPROC started by Rule Management Services (RMS). You can establish a
separate queue for NCL processes that perform the same task. For example, you
might have one queue called RESTARTTERM for executing NCL processes that restart
a failed Pathway terminal. When the START verb executes an NCL process that
performs this task, the START verb sends the NCL processes to RESTARTTERM.
You can limit the number of NCL processes that can execute concurrently in a queue.
For example, you might allow only ten NCL processes to execute in RESTARTTERM
because, if many Pathway terminals fail simultaneously, you do want to flood the
system with NCL processes that try to simultaneously restart all terminals. For the
first ten failures, an NCL process starts recovery immediately; one NCL process for
each failure. However, if an eleventh Pathway terminal fails while the ten NCL
processes are still executing, the eleventh NCL process is queued (in the NCL queue
file) until one of the ten NCL processes stops executing.
Figure 5-1 is a simplified view of NCL queues that shows how NCL queues are used to
control the number of concurrently executing NCL processes. The figure shows a
single NonStop NET/MASTER MS application process in which a single NCL
processing is executing. The figure shows three NCL queues, for which the execution
limits are one, three, and five NCL processes, respectively. The figure shows that the
application process stores delayed or queued NCL processes in an NCL queue file
called pQnnn. (Another file (pQnnnA) is also used by the application process to store
NCL queue information, and is shown in the figure. (An NCL queue recovery file is
not used by persistent application processes because these processes do not run
persistent NCL processes.))