Health metric rules have two levels. First you specify the cluster member for which a set of rules will apply. Then you configure the individual rules. You can use either of the following as the basis of the rule:
• A characteristic of the specified member’s child cluster members: either the number that is active or inactive, or their health level.See Understanding Health Metric Rules for more details.
Put the most severe health level first Within each clustermember element, the MM server examines the setproperty element that is closest to the top of the file first. When a health metric threshold for a cluster member is met, the application stops and does not process additional thresholds for that member. If you are configuring all three levels, you would put critical first, then warning, and finally normal.See Health Metric Rule Configuration Reference for more details on the settings referenced in this procedure.
As with any procedure that changes files, ensure the emonitor project is backed up before you edit its CDD file.
1.
2. Open the MM.cdd in the Cluster Deployment Descriptor editor and select the Agent Classes tab
3.
4. Click Add. You see a configuration panel.
7. Click Add. You see fields to define a health metric rule for this cluster member.
8. You can configure health metric rules in two ways. See Two Types of Thresholds for advice on which method to use. Go to one of the following procedures to continue, depending on how you want to configure this rule:This procedure flows from the procedure To Set Up a Health Metric Rule. See Health Metric Rule Configuration Reference for more details on the properties referenced in this procedure.
3. In the Condition Type field, select Child Cluster Member.
4. In the Path field, enter the cluster path of the child cluster member you want to use to compute this health level metric. For example, in computing cluster health you might specify cluster/machine, and for machine health, you might specify cluster/machine/process.
5. In the Threshold field enter the threshold percentage. If the percentage of child cluster members that match the criteria specified meets or exceeds this threshold, then the health level of the parent cluster member is set to the Health Level field value.
6. In the Properties sheet add one or both of the following properties to set the criteria for counting child members:
− To use the number of active or inactive child members, add a property called active. Set the value to false to count only inactive members. Set the value to true to count only active members.
− To use the number of child members at a certain health level, add a property called healthLevel and set the value to one of critical, warning, or normal.
7. Repeat this procedure to configure this cluster member’s thresholds for remaining health levels (critical, warning, or normal), as desired.
8.
9. If you have finished configuration, start the emonitor project using the modified CDD file. This starts the MM server. See Starting MM Server and Logging On to the Console.This procedure flows from the procedure To Set Up a Health Metric Rule. See Health Metric Rule Configuration Reference for more details on the properties referenced in this procedure.
3. In the Condition Type field, select Notification.
4. In the Range field enter a time period in milliseconds. Or enter 0 to specify no time period. If the Threshold number of alerts (of the severity value) is received in the Range period, the health metric is set to the Health Level value. If you specify zero (0) then the health level is changed after receiving the Threshold number without regard to any time window.
6. In the Properties sheet add a property called severity. As the value specify one of critical, warning, or normal.
7.
8. If you have finished configuration, start the emonitor project using the modified CDD file. This starts the MM server. See Starting MM Server and Logging On to the Console.
Copyright © TIBCO Software Inc. All Rights Reserved.