3.2.9 Setting up the monitoring items for service performance
You must set up monitoring items for each monitored service.
- Organization of this subsection
(1) Before you start
Verify that you have the service group administrator permissions.
Verify that the monitored service has been registered.
For details about how to register monitored services, see 3.2.2 Registering monitored services.
-
If you monitor Web transactions, verify that the Web transactions have been registered.
For details about how to register Web transactions, see 3.2.5 Setting up the Web transactions to be monitored.
-
When setting the following monitor items, check that monitoring of the monitored service to set the monitor item for is stopped.
-
Whether SLO monitoring has been implemented
-
Whether to monitor trends
-
Whether predictive error detection has been implemented
-
Period of analysis of predictive error detection
For details about how to stop monitoring, see 4.2.2 Stopping monitoring.
-
(2) Procedure
The following shows the Settings window used in this task:
To set up monitoring items for service performance:
Click the Settings button.
In the Setting menu area, select Monitor settings.
The Monitor settings area is displayed.
From the Services area, select a monitored target of a monitored service.
When you select a monitored target of a monitored service, the service group name, monitored service name, and monitored target are displayed in the Monitor settings area. The current values are displayed under SLO monitor settings and Error Predict. settings. Immediately after a monitored service has been registered, the default values are set.
If you will be running threshold value monitoring or trend monitoring, select the Item name check boxes under SLO monitor settings for the items that you want to monitor, and then enter values in Threshold.
An error message is displayed if an Item name check box is selected but no value is specified for that item or an invalid value is entered in the text box.
If you will be running trend monitoring, select the Trend monitor check boxes for the items that you want to monitor under SLO monitor settings, and then enter the reference time for trend calculation.
The Trend monitor check boxes are enabled only when Item name check boxes are selected. In the Trend monitor text box, enter the time to be subject to trend monitoring.
An error message is displayed if a Trend monitor check box is selected but no value is specified for that item or an invalid value is entered in the text box. Note that there is no check box for Error rate, because trend monitoring is not applicable to error rate.
Under Error Predict. settings, enter appropriate values in Days in baseline calculation and Days till start.
An error message is displayed if an invalid value or nothing is entered in a text box. If you will not be performing out-of-range value detection, leave the default values in Days in baseline calculation and Days till start.
-
Select Period of analysis in Error Predict. Settings.
Select the period of analysis for out-of-range value detection from the following items in the pull-down menu.
Short range: Past 60 days
Long range: Past 5 years
If you will be performing out-of-range value detection, select the Item name check boxes for the items that you want to monitor under Error Predict. settings and then select their Sensitivity settings.
Select an item that you want to monitor, and then select High, Middle, or Low as its sensitivity. As the sensitivity becomes higher, it becomes easier to detect the item. As the sensitivity becomes lower, it becomes harder to detect the item. Initially, set the sensitivity to Middle, and then you can adjust it later as needed after checking the number of items detected.
If you perform out-of-range value detection with multiple monitoring items combined, select Throughput from the Correlated items pull-down menu on the Avg. response row under Error Predict. settings.
Click the Apply button.
If the monitoring items have been set up successfully, a dialog box to that effect is displayed.
When you click the OK button in the dialog box, the settings are applied.
(3) Supplementary information
-
As setting the Period of analysis in Error Predict. settings to Long range will lower the accuracy of predictive error detection, we recommend changing the outlierRate property in the system definition file (jp1itslm.properties) to adjust how often warning events occur. However, as changing the outlierRate property will affect the entire system, the change of the degree of occurrences of warning events will be constantly applied. For example, when the forms of the graph greatly differ between busy periods and normal times other than the busy periods, changing the outlierRate property to suppress warning events during the busy periods might cause warning events to be suppressed during normal times. For this reason, to avoid affecting operation during normal times, it is necessary to take measures such as ending SLM - Manager before the busy periods, and restarting SLM - Manager after changing the outlierRate property.
For details about the system definition file (jp1itslm.properties), see 5.6 Editing the system definition files to change settings.
-
In the case of a Web transaction, the Period of analysis in Error Predict. settings cannot be specified. Follow the specification of All Web Access.
(4) Next task
-
3.2.10 Setting up the monitoring items for system performance (working with Performance Management)
4.2.1 Starting monitoring (when not working with Performance Management)