Advanced resource monitoring

Many predefined conditions and responses are provided by the various resource managers on your system. These predefined conditions and responses are provided as an administrative convenience.

As described in Basic resource monitoring, many predefined conditions and responses are provided by the various resource managers on your system. These predefined conditions and responses are provided as an administrative convenience. As described in Creating a condition/response association, you can use them to create condition/response associations for monitoring. However, the predefined conditions and responses may not always meet your needs. In such cases, you can use advanced resource monitoring. Advanced resource monitoring includes:
  • Creating your own conditions that can then be linked with one or more responses and monitored by RMC. If the condition you want to monitor is similar to one of the predefined conditions available on your system, you can copy the existing condition and modify it as needed. If none of the existing conditions are similar to the condition you want to monitor, you can create a new condition from scratch. This involves identifying the attribute you want to monitor for one or more resources of a particular resource class. Since persistent attributes are generally unchanging, you will usually monitor a dynamic attribute. If none of the dynamic attributes provided by the resource managers contains the value you want to monitor, you can create a sensor—a command to be run by RMC to retrieve the value you want to monitor. For more information, see Creating, modifying and removing conditions.
  • Creating your own compound conditions that can then be linked with one or more responses and monitored by RMC. If the compound condition you want to monitor is similar to one that already exists on your system, you can copy the existing compound condition and modify it as needed. If none of the existing compound conditions is similar to the compound condition you want to monitor, you can create a compound condition from scratch. This involves identifying the existing conditions whose event notifications you want to monitor, correlate, and evaluate.
  • Creating your own responses that can then be linked with conditions. You can use predefined response scripts in your responses and you can also create your own response scripts. For more information, see Creating, modifying, and removing responses.

Once you know how to create conditions, compound conditions, and responses, be aware that, to be effective, you need to link the conditions or compound conditions together with one or more responses and start monitoring. These tasks are described in Creating a condition/response association and Starting condition monitoring. For detailed syntax information on any of the commands, see Technical Reference: RSCT for AIX® and Technical Reference: RSCT for Multiplatforms.

Tuning reporting intervals

You can adjust reporting intervals for resource dynamic attributes. A reporting interval is the amount of time between a resource manager's sampling of values. You can set reporting intervals for all classes, a single class, or a single attribute of a class.

To adjust the reporting intervals for resource dynamic attributes that are defined in a class definition file, you can create a reporting interval override file. This file is called:

/var/ct/cfg/ctrmc.rio
Information in this file only applies to resource dynamic attributes that have reporting intervals, which have a variable type of Counter or Quantity.
ctrmc.rio is a plain text file, in which a line in the file contains two white-space-separated tokens, as follows:

class_name[:attr_namereporting_interval
The first token is a class name as returned by the lsrsrc command, or, a class name followed immediately by a colon, which is followed by an optional resource dynamic attribute name as returned by the lsrsrcdef command. If only a class name is specified, the reporting interval applies to all resource dynamic attributes of the class that have reporting intervals. If a class name and attribute name are specified, the reporting interval applies to the named attribute only, if appropriate. If the attribute does not have a reporting interval, the line is ignored. The second token, the reporting interval, is an unsigned integral value that is interpreted as a number of seconds. If the class name, without any attribute names, is the keyword ALL, the reporting interval applies to all dynamic resource attributes of all resource classes. The last specification for a given attribute applies.
In the following example:

Foo:larry       10
Foo             15
Foo:curly       20
the reporting interval would be 15 for larry and 20 for curly. All other dynamic attributes of the Foo class would have a reporting interval of 15. Blank lines, lines that begin with the number sign ( # ), and lines that contain unrecognized tokens are ignored.

The reporting interval override file is read by the RMC daemon upon start. If the refresh -s ctrmc command is invoked, the file is re-read. In this case, any new reporting intervals apply only to future event registrations.