Administration Guide

Chapter 26. Using the Problem Management subsystem

The Problem Management subsystem (pman) provides an infrastructure for recognizing and acting on problem events in your SP system. This infrastructure is based on an Event Management application that provides configurable access to Event Management client and resource monitor function without the necessity of writing C programs that use the Event Management APIs.

For more information, see the book RS/6000 SP High Availability Infrastructure.

Note:

|Enabling restricted root access and setting authorization for AIX |remote commands to none might limit Problem Management. See Authorizing event response actions. |

To understand the information presented in this chapter, you need to be familiar with Event Management concepts and terminology. The following are some important Event Management terms:

event

In Event Management, the notification that an expression evaluated to true. This evaluation occurs each time an instance of a resource variable is observed.

expression

In Event Management, the relational expression between a resource variable and other elements (such as constants or the previous value of an instance of the variable) that, when true, generates an event. An example of an expression is X < 10 where X represents the resource variable IBM.PSSP.aixos.PagSp.%totalfree (the percentage of total free paging space). When the expression is true, that is, when the total free paging space is observed to be less than 10%, the Event Management subsystem generates an event to notify the appropriate application.

instance vector

An obsolete term replaced with the term resource identifier.

predicate

An obsolete term replaced with the term expression.

rearm expression

In Event Management, an expression used to generate an event that alternates with an original event expression in the following way: the event expression is used until it is true, then the rearm expression is used until it is true, then the event expression is used, and so on. The rearm expression is commonly the inverse of the event expression (for example, a resource variable is on or off). It can also be used with the event expression to define an upper and lower boundary for a condition of interest.

rearm predicate

An obsolete term replaced with the term rearm expression.

resource

In Event Management, an entity in the system that provides a set of services. Examples of resources include hardware entities such as processors, disk drives, memory, and adapters, and software entities such as database applications, processes, and file systems. Each resource in the system has one or more attributes that define the state of the resource.

resource identifier

In Event Management, a set of elements, where each element is a name/value pair of the form name=value, whose values uniquely identify the copy of the resource (and by extension, the copy of the resource variable) in the system.

resource monitor

A program that supplies information about the resources in the system. It can be a command, a daemon, or part of an application or subsystem that manages any type of system resource.

resource variable

In Event Management, the representation of an attribute of a resource. An example of a resource variable is IBM.AIX.PagSp.%totalfree, which represents the percentage of total free paging space. IBM.AIX.PagSp specifies the resource name and %totalfree specifies the resource attribute.

For more information on Event Management services, see the book RSCT: Event Management Programming Guide and Reference.

The major components of the Problem Management subsystem are:

• A Problem Management daemon (pmand) that provides access to events generated by Event Management on all nodes. In addition, the following function is provided to work in conjunction with pmand:
  • A command, pmandef, that provides for subscribing to Event Management events and associating actions for those events
  • A command, pmanquery, that queries the SDR for a description of a Problem Management subscription
  • A script, pmandefaults, that causes pmand to register for a set of default events
  • A script, notify_event, that mails an event notification when an event occurs
  • A script, log_event, that logs a record of an event to a regular wraparound file
• A resource monitor daemon, pmanrmd, that provides resource variables to Event Management. The following function is provided to work in conjunction with pmanrmd:
  • Sixteen resource variables. The resource variables are named IBM.PSSP.pm.User_state1 through IBM.PSSP.pm.User_state16.
  • A command, pmanrminput, that provides for associating values with the sixteen supplied resource variables.
  • A sample file, pmanrmd.conf, for configuring the pmanrmd daemon.
  • A command, pmanrmdloadSDR, for loading the configuration information into the System Data Repository.

This chapter provides information on each of the components of Problem Management.