An action is a predefined, reusable, automated response to an event. You can use actions to automate your runbooks.
A policy can trigger an action in response to any event. You configure which actions are triggered by which events when you configure policies.
You can create these types of actions:
Not all actions are applicable to all application environments or to all situations. Below are some general guidelines concerning different types of actions. For more details, see the pages on the specific actions before you assign an action to a policy.
The Controller limits the actions invoked based on the number of triggering events per event type. There is a maximum of ten events for any single event type that can trigger actions in a given minute. If the number of triggering events per type exceeds the limit, the actions that would have been triggered by the excess events are not started. You will not see a visual indication that these actions are not being started.
For example, your application can have up to ten Health Violation Started events triggering actions and up to ten Resource Pool Limit Reached events triggering actions within the same minute. But if you have eleven Health Violation Started events firing, the action that would be triggered by the eleventh event is not started.
To reduce unnecessary actions, there is a limit on the number of diagnostic and remediation actions that will invoke. The default limit is five actions per minute per machine for each type of action.
If, for example, a policy is configured on all the nodes where there are 100 nodes triggering actions, randomly selects five of the actions to execute.
To avoid exceeding the limits, design your policies so that they do not trigger an excessive number of actions for any particular event. You can generate fewer events by configuring the affected entities of your health rules at the tier level. See entities affected by a health rule and by health rule types in Health Rules.
For actions that take thread dumps or run a local script, you can optionally require email approval to run the action whenever it is triggered. If you check this option, human intervention is required before the automated action actually starts.
If you specify the approval required option when you configure the action, when the action is triggered an email containing a link is sent to the configured email address. The link presents a login screen (if the user is not already logged in to ) and after the user logs in, a dialog requesting approval to take the thread dump or run the script. The user can click in this dialog to approve and start the action or cancel the action.
If you do not check the Require Approval option before executing the Action checkbox, the action will start automatically with no human intervention.
To configure actions, you need the Configure Actions permission. |
To access the Actions configuration panels:
While actions are being triggered by events, you can get a summary of their execution in the Actions column in the Events list. You can access the Events list from the Events tab of the application, tier, or node dashboard.
The icon type indicates action type; an icon in the Actions column indicates that an action was triggered for this event. If the action icon is greyed out, the action is either still executing or failed to complete. If the icon fully displays, then the action executed successfully.
You can prevent policies from firing actions for a configurable time period.