Investigating data flow run failures
From the Data Flow landing page, you can access detailed reports on any errors that occur while a data flow is being processed. By analyzing these the error reports, you can quickly diagnose the root cause of an error.
Understanding error messages
Each error message provides information about the data flow run in which the error occurred and a link to a stack trace. You can review the method calls in the stack trace to understand the point at which your app encounters an error and which exceptions are raised.
- In the header of Dev Studio, click .
- Click the appropriate tab for the type of data flow run that you are
investigating.
- Batch processing
- Real-time processing
- Single case processing
- In the list of runs, open the failed data flow run by clicking its name.In the following example, the data flow run failed because the number of records in which failures occurred exceeded the threshold:
An error message for a failed data flow run - In the error message for the failed run, click
Details.
Result: The stack trace is displayed in a new window, with additional information about the error, as in the following example: com.pega.dsm.dnode.impl.dataflow.resilience.policy.MaxFailedRecordsPerRunPolicy$ErrorThresholdReachedException: The run failed, because it exceeds the maximum number of failed records, which is currently set to 1
- Optional: To improve the readability of the stack trace, right-click inside the error window, and then click View page source.
Viewing failure counts across components
You can view the number of failed records for each component in a data flow to detect which components are the source of errors.
- On the data flow run tab, click the Component statistics tab.
- Review the component statistics to understand which data flow component registered failures.
- For a component in which errors occurred, click the number in the
Failed records column.In the following example, all records failed at the filter component:
Component statistics Result: A report is displayed, showing the exceptions that the system encountered while running the data flow and processing records.
Viewing failure counts across nodes
If you encounter an error, you can view on which node the error occurred. You can also view the throughput for each node, which might give an indication as to which node is the slowest.
- On the data flow run tab, click the Distribution details tab.
- Review the node statistics to understand which node registered failures.
- For a node on which errors occurred and which you want to inspect, click the
number in the Failures column.In the following example, errors occurred on both nodes that processed records in the data flow run:
Distribution statistics Result: A report is displayed, showing the exceptions that the system encountered while running the data flow and processing records.
Viewing exceptions
You have identified the components and nodes on which failures occurred during the data flow run. You can now examine the exceptions that were thrown when your app processed each record that failed.
- In the Show last preserved data flow errors window, review
the exceptions.
For each record, the report provides information about the exception, such as the data flow stage at which the exception was thrown and the node that processed the record:
Report with exceptions - Choose a record that you want to inspect, and then click View
details.
Result: The stack trace lists the method calls that were in progress when an exception was thrown. This information can help you understand at which point your application ran the error occurred: Exception details - In the Show last preserved data flow errors window, for a
record that you want to inspect, click View input
record.
Result: Details about the record that caused the error are displayed: Input record
Previous topic Troubleshooting data flows Next topic Data flows cannot parse JSON records