PSL discovery programming capsule

This section  summarizes typical discovery programming. Use a PSL discovery script to define the PATROL Agent namespace variables for each detected application instance, detect application instances and create icons for them, set the current operational status of each instantiated application instance and check whether previously discovered application instances still exist on the monitored machine and destroy any obsolete application instances.

Typical PSL discovery algorithms

The following table lists typical discovery algorithms.

Typical PSL discovery algorithms

PSL discovery programming guidelines

Discovery should be built to run as efficiently as possible using the following guidelines:

Application instance detections

• Define the environment for a new instance only at instance creation (i.e. when the create() function is called).
• You should create an icon for every detected application instance you want your users to see.
• Use the Create Icon for Class option only if more than one instance of an application class can occur.

Process cache

• Restrict the process list to a local variable containing all processes related to all instances of an application versus repeated process() function calls.
• Check, using full_discovery(), to determine whether the process list for an application should be rebuilt. If discovery builds a list of process IDs for an application instance, populate the PID list once at instance creation. Then, rebuild the list on full discovery only if the instance state is other than OK. Use the PSL process() function to obtain a list of processes from the Determine and set state only during the application check cycle.
• Determine and set state only during the application check cycle

Parameters

• Avoid setting parameter values in discovery scripts.

Other guidelines

• Discovery scripts should be the only place that the PSL create(), destroy() and change_state() functions are used.
• Use the default discovery cycle unless you have good reason to change it.

Typical PSL discovery script structure

A typical PSL discovery script has the following sections:

• Initialization
• Application instance detection
• Application instantiation
• Application instance destruction

The following table describes of each section and its usage.

Typical PSL discovery script sections

If the application instance cannot be discovered and instantiated without user interaction for object-specific authentication, define a separate 'CLASSNAME_SETUP' application class to collect this information by menu commands or setup parameters and store it in the PATROL Agent namespace.

For ease of programming, combine application detection and instantiation so that the parsing of information collected during application detection is combined with the creation of objects and the setting of their status and instance attributes.

Depending on the managed application class, you may wish to maintain audit records on destroyed application instances.

Prototype PSL discovery script

This example illustrates the concept of unconditional instantiation. In unconditional instantiation, no logic is provided to determine whether or not the application instance should be created. 
The following discovery script does these things:

• Checks to see if the application instance ck.process of the application class ck.process already exists.
• If the ck.process does not yet exist, the PSL create() function creates an instance of the application class ck.process and sets its state to "OK."
• If the application instance ck.process already exists, nothing happens.

PslDebug = 0;
if(!exists("/ck.process/ck.process")){
    create("/ck.process/ck.process","ck.process",OK);
}

PSL functions used in PSL discovery

The following table lists functions that are often used in discovery scripts.

Common PSL discovery functions

 Variables used in PSL discovery

The following table lists variables that are often used in discovery scripts.

Common PSL discovery variables