On-demand synthetic monitor executions for CI/CD

In addition to scheduling synthetic monitor executions at regular intervals (in Frequency and locations settings), you can also trigger on-demand executions for browser as well as HTTP monitors from assigned public and private locations. You can even trigger on-demand executions of autocreated HTTP monitors for credential synchronization. In the Dynatrace web UI, select On-demand execution at the top of a monitor's details page.

You can also use the powerful Synthetic - On-demand monitor executions API to trigger executions of multiple monitors and retrieve the results. On-demand execution requests take precedence over your scheduled executions and are sent to the top of the request queue.

During synthetic monitor creation and in edit mode, you can also set up a monitor to be executed on demand only—select the On demand only frequency.

On-demand executions, with their powerful features, are a valuable tool in your CI/CD strategy:

• Monitors are executed on demand from assigned public and private locations.
• On-demand execution results may be included or excluded from overall monitor results.
• Executions may be included or excluded from problem detection.
• They can be triggered via the web UI or API.

These executions can validate that a new deployment of your software is successful and can be integrated with build tools like Jenkins or keptn.

On-demand executions are also very valuable as quick monitor tests, for example, you might test a monitor after developing a complex script. Likewise, such executions can verify that a fix for a Dynatrace-detected problem was effective, for example, if you added host memory or reverted to an older application version.

With their ability to be included or excluded from overall monitor results and problem detection, on-demand executions from assigned locations offer many more options than local playback of browser clickpaths.

UI-based on-demand executions

The On-demand execution button on synthetic monitor details pages enables you to trigger executions for the selected monitor from all or a selected assigned location. It also displays a list of all on-demand executions for the monitor over the preceding six hours, triggered by any user via the web UI or API.

Trigger an execution

Any user with the Access environment permission at the environment or management-zone level can trigger an execution via the UI.

Even if you disable scheduled synthetic monitor execution during maintenance windows, you can still execute monitors on demand.

1. Select On-demand execution at the top of the selected monitor's details page.

2. You are directed to the On-demand execution page. Select a specific location or all Locations—you can only select from public and private locations assigned to the monitor.

3. Select a Processing mode—note that this applies to all executions if you've selected all locations.

   • Standard (default)—the executions contribute to problem detection and availability and performance statistics. For browser monitors, you can view data points on the Multidimensional analysis page. For HTTP monitors, you can access execution results when you Analyze execution details.
   • Disable problem detection—the executions contribute to availability and performance statistics but not to problem detection. For browser monitors, data points appear on the Multidimensional analysis page; for HTTP monitors, you can view execution results when you Analyze execution details.
   • Execution details only—the executions do not contribute to availability and performance statistics or to problem detection. For HTTP monitors, you can still access results when you Analyze execution details; for browser monitors, these data points appear on the Multidimensional analysis page.

   Note that for all processing modes, executions are visible in the list of on-demand executions in the web UI and retrieved via the on-demand executions API. Summary and detailed execution information is available for six hours via the API.

4. Fail on performance threshold violation—by default, on-demand executions fail when they violate a performance threshold, given that their main purpose is to validate new software versions in your CI/CD pipeline. However, you can disable this setting so that on-demand executions behave like regularly scheduled executions (in other words, performance threshold violations will not cause monitors to fail).

5. Select Trigger now—you see a summary dialog listing the execution locations; the execution list displays new entries for the triggered executions. The Execution stage initially is Triggered.

   Note that executions might not all begin at the same time from different locations—executions might take longer to begin from public locations than from private locations. See additional information on throttling below.

   

Throttling

• There is a mandatory gap per user of 60 seconds between consecutive on-demand exections of a monitor from the same location (whether triggered via the web UI or API).

  Examples

  All these examples assume execution from the same location.

  • User A triggers a monitor via API within 60 seconds of triggering the same monitor via UI—throttled.
  • User B triggers a monitor via API within 60 seconds of user A triggering the same monitor via UI—no throttling.
  • User B triggers a monitor via UI within 60 seconds of user A triggering the same monitor via UI—no throttling.
  • User B triggers a monitor via API within 60 seconds of user A triggering the same monitor via API—no throttling.

• When triggering multiple on-demand executions via the API, there is a limit of 100 executions per batch.

Failure to be triggered

On-demand executions might not be successfully triggered for various reasons, for example, when a monitor is disabled, a location is down, or when throttling is in effect for a monitor on a given location.

There are several additional reasons that on-demand executions might not be successfully triggered via the API: monitor deletion, incorrect specification of monitor or location IDs, incorrect specification of related service or application IDs, location deletion, problems with the public Synthetic infrastructure, or problems with your Dynatrace monitoring environment.

If executions could not be successfully triggered via the UI, the reasons are shown in the Triggering status summary after you select Trigger now. Details of executions not triggered via the API are returned in the triggeringProblemsCount and triggeringProblemsDetails response parameters for the POST request.

Note that executions that could not be triggered are different from executions that were triggered but could not be executed—see the API section below for information.

List of executions

The list of executions shows all on-demand executions (triggered by any user via the web UI or API) for a given monitor within the last six hours.

• The Execution ID is a unique ID assigned to each execution; if you trigger executions from all locations, each location's execution has a different ID.
• The Triggered column shows the start time of the execution in the logged-in user's time zone.
• Source shows whether the execution was triggered via the web UI or the API.
• The User column shows the Dynatrace user ID of the user who triggered the execution.
• The Location column shows the name of the public or private location from which the monitor was executed.
• Execution stage identifies the different stages of the on-demand execution. The initial value is Triggered. When an execution is complete, the value changes to Executed. Basic results such as duration and HTTP status code are available at this stage. The progress spinner continues to be displayed at the Executed stage until detailed results are available and the value changes to Data retrieved.

• The Result column displays whether the execution was a Success or Failure (with an accompanying Failure reason).
• Select the expand button in the Details column to view execution duration and a link to results.

Results

When the Execution stage in the list of executions changes to Data retrieved, you can follow the execution details link to view detailed results for the browser or HTTP monitor that was executed. Note that this link is not available if you opt to exclude the execution from metric calculations (Execution details only).

For browser monitors, you're directed to the Multidimensional analysis page with the data point selected in the scatter plot. On-demand executions are identifiable by shape in the scatter plot as well as by an annotation in the list of data points.

For HTTP monitors, you're directed to the On-demand execution tab of the Analyze execution details page. You can also access this tab by selecting Analyze execution details from the HTTP monitor details page. Select an on-demand Execution from the dropdown to view its details.

API: Synthetic - On-demand monitor executions

Check the section on throttling above for on-demand execution limits via the web UI and/or API.

API-based on-demand executions offer greater flexibility and scalability than the UI. The Synthetic - On-demand monitor executions API offers:

• Retrieval of the list of on-demand executions (triggered via the web UI or API by all users in your environment) in the preceding six hours. The list includes execution IDs for each execution. You can filter this list by:

  • Execution and data-delivery timestamps
  • Batch IDs
  • Monitor IDs
  • Location IDs
  • User IDs

• The ability to execute multiple monitors on demand (batch execution) by specifying monitor IDs, and, optionally, location IDs. If you don't specify locations, a monitor is executed from all assigned locations. In the POST request for triggering on-demand executions, each specified monitor is executed from its respective specified locations (or all assigned locations if none are specified).

• The ability to execute a group of monitors on demand by specifying common tags and/or IDs of related services or applications. Note that all specified conditions must match for a monitor to be executed on demand.

  If you specify three monitors by ID and one tag in your POST request, each of the three monitors plus all monitors matching the tag will be executed.

• Automatic assignment of a batch ID for all executions of the POST method, whether for a single monitor or multiple monitors. This enables the retrieval of results by batch ID as well.

• The ability to stop triggering all executions in a batch if there is a problem with triggering any execution (stopOnProblem parameter), for example, if the monitor ID is incorrect/missing or the monitor is deleted.

• Retrieval of the basic as well as more detailed results of an execution (for which you provide an execution ID).

  Basic results are best suited for CI/CD purposes and include the number of requests/events executed, success/failure result, and some key metrics. For example, basic results for HTTP monitors report total request size (for all requests taken together), time to first byte, TLS handshake time, TCP connect time, DNS lookup time, and the final HTTP status code.

  Detailed results are more suitable for troubleshooting. For HTTP monitors, this is the entire set of results visible in the web UI when you opt to Analyze execution details.

• Batch retrieval of results when you provide a batch ID.

  Batch results include information on executions that were triggered and those that resulted in an outcome.

  Note that the executedCount is the number of executions that resulted in success or failure. failedToExecuteCount is the number of executions that were triggered but for which results are not available for technical reasons (such as a problem with the ActiveGate or Synthetic engine or results that could not be sent to the Dynatrace Cluster). executedCount + failedToExecuteCount = triggeredCount.

• Granular permissions for triggering on-demand executions and retrieving data.

  To trigger executions (POST), you need rights to create monitors (ExternalSyntheticIntegration token scope) or the syntheticExecutions.write token scope, which enables you to trigger executions but not create new monitors.

  To retrieve data (GET), you need any of the ExternalSyntheticIntegration, ReadSyntheticData, or syntheticExecutions.read token scopes.