Administration > Monitor Administration > SiteScope Data Collector Configuration > Managing SiteScope > SiteScope Health > Monitoring SiteScope Server Health > Understanding SiteScope Health Monitoring
Understanding SiteScope Health Monitoring
This section contains information about how to interpret the results of SiteScope Health Monitoring and some actions to take if errors are detected. See SiteScope Health Monitor Reference for more details on the specific SiteScope Health monitor measurements.
SiteScope Log Events
The Log Event Monitor is the equivalent of a SiteScope monitor group that watches the SiteScope Error Log (error.log) for certain events. These events include Log entries indicating that a monitor has been "skipped" or there was a problem in reporting data to another application.
A SiteScope monitor will be reported as "skipped" if the monitor fails to complete its actions before it is scheduled to run again. This can occur with monitors that have complex actions to perform, such as querying databases, stepping through multi-page URL sequences, waiting for scripts to run, or waiting for an application that has hung. This can also happen if there are too many monitors waiting to run that require a process from the process pool.
For example, assume you have a URL Sequence Monitor that is configured to transit a series of eight Web pages. This sequence includes performing a search which may have a slow response time. The monitor is set to run once every 60 seconds. When the system is responding well, the monitor can run to completion in 45 seconds. However, at times, the search request takes longer and then it takes up to 90 seconds to complete the transaction. In this case, the monitor will not have completed before SiteScope is scheduled to run the monitor again. SiteScope will detect this and make a log event in the SiteScope error log. The SiteScope Log Event Monitor will detect this signal an error status.
A monitor may also skip if it is a monitor type that requires a process from the process pool but the process pool limit has been reached. Generally, this is not likely to happen but may occur in some situations with high monitoring load. The SiteScope Health Log Event Monitor also watches for process pool events.
Skipped monitors cause a number of problems. One is the loss of data when a monitor run is suspended due because a previous run has not completed or has become hung by a unresponsive application. Skipped monitors can also cause SiteScope to automatically stop and restart itself, an event that is also monitored by the SiteScope Health Log Event Monitor. A restart is done in an effort to clear problems and reset monitors. However, this can also lead to gaps in monitoring coverage and data. Adjusting the run frequency (Frequency) at which a monitor is set to run or specifying an applicable timeout value can often correct the problem of skipping monitors. Investigation of unresponsive systems that are being monitored may also be necessary.
Note: A Max Monitor Skipping setting has been added to allow monitors that are skipping to be disabled automatically. If this occurs, SiteScope is not restarted but an e-mail is sent to the SiteScope administrator about the skipping monitor to signal the disable event. This optional functionality is disabled by default but can be enabled by changing the _shutdownOnSkips to remove the value in the master.config file or remove the setting entirely.
Note: A setting for controlling the maximum number of processes available is available in the master.config file. The default is _processPoolMaxPerPool=50. You should only change this setting if adjustments to monitor configurations do not resolve the monitor performance problems.
The Log Event Monitor is also configured to report log events that indicate a problem with the transfer of SiteScope monitor and configuration data to a Mercury Business Availability Center installation. See the section on Integration with Mercury Business Availability Center for more information on Troubleshooting Data Reporting to Business Availability Center.
SiteScope Monitor Load
The Monitor Load Monitor is the equivalent of a SiteScope monitor group that watches how many monitors are running and how many are waiting to be run. Watching monitor load is important to help maintain monitoring performance and continuity. If the number monitors waiting approaches or exceeds the number of monitors running, adjustments should be made to monitor configurations to reduce the number of monitors waiting to run. Generally, this can be done by reducing the run frequency of some monitors.
SiteScope Server Health
The Health of SiteScope Server Monitor is the equivalent of a SiteScope monitor group that monitors server resources on the server where SiteScope is running. This includes monitors for CPU, disk space, memory, and key processes. A problem with resource usage on the SiteScope server may be caused by monitors with configuration problems or may simply indicate that a particular SiteScope is reaching it performance capacity. For example, high CPU usage by SiteScope may indicate that the total number of monitors being run is reaching a limit. High disk space usage may indicate that the SiteScope monitor data logs are about to exceed the capacity of the local disk drives (see the section on Log Preferences for SiteScope data logging options).
| Copyright 2006 Mercury Interactive Corporation |