Table of Contents
Advertisement
Detailed Description
The Detailed Picture of Hardware Monitoring
EMS Hardware Event Monitor
The EMS hardware event monitor is the key component in the event monitoring architecture. An event
monitor is a daemon process, running in the background continuously. The event monitor watches all
instances of the hardware resources it supports, waiting for the occurrence of any failures or other unusual
events. The monitor may use polling, asynchronous event detection, or both.
When an event occurs, the monitor alerts EMS and passes it the appropriate event message. The event
monitor also tells the PSM about the event. If the event is serious enough the PSM will change the status of
the hardware to DOWN.
Two configuration files control the operation of each hardware event monitor:
•
Global monitor configuration file. The settings defined in this file are used for all hardware event
monitors, unless overridden by a monitor-specific file.
•
Monitor-specific configuration file. Each monitor includes its own configuration file with optimized
settings. The settings defined in the monitor-specific file override corresponding settings defined in the
global configuration file.
NOTE
The settings defined by the monitor-specific configuration file have been carefully selected to
meet the needs of most users. It is possible to alter these settings, but it is not recommended
unless you fully understand the implications of doing so. For information on modifying the
monitor-specific configuration files, see Chapter 5, "Hardware Monitor Configuration Files."
NOTE
As of the June 2000 release, several of the hardware monitors have been converted to be
"multiple-view" (Predictive-enabled). These monitors use a different file for configuration, the
Client Configuration File.
Polling or Asynchronous?
Hardware event monitors employ two methods of tracking events: polling and asynchronous event detection.
A monitor may use one or both of the methods to detect events.
Using polling, a monitor checks the status of its hardware resources at regular intervals, typically 60
minutes. Any unusual condition reported by the hardware will trigger an event by the monitor. The polling
interval is selected to provide reasonable detection without impacting system performance. The main
disadvantage of polling is that an event will not be detected until the next time the resource is polled, which
makes the system vulnerable to another hardware failure.
Asynchronous detection allows a monitor to detect an event when it occurs, usually during an I/O to the
device. An event typically results in a log entry made by the hardware device driver. The monitor detects the
log entry and initiates the event notification. Asynchronous event monitoring allows immediate notification
and response to a critical situation.
Startup Client
The startup client launches and configures the hardware event monitors each time the system is started, or
following the execution of the IOSCAN utility (thus performing a real/hard ioscan). The startup client starts
each monitor and configures its hardware resources using a set of default monitoring requests.
62
Chapter 3
Advertisement
Table of Contents
