Troubleshooting errors and failures in a failed process instance

Using the Process Inspector, you can take corrective actions on process instance errors one at a time or in bulk across multiple locations. In most cases, you can either try to run the process again, edit data and then try again, or skip a failed step. The best method to use depends on the nature of the failure.

About this task

Failures can be of two types, system failures or failures caused by invalid or corrupted data. A process instance might fail because the network connection is down or a power failure occurs. Or it might fail because changes occur in the execution environment after the instance is migrated, and those changes cause data to become invalid or corrupted.

To recover from a system failure, use the Process Inspector in the Process Admin Console web interface. First, make sure that the condition that caused the failure is corrected. Then, to view and process the errors, select Retry failed steps. Retry failed steps resumes or restarts failed process instances.

If an activity failed to complete because of invalid data, use the Process Inspector to view and resolve those errors. Click Edit data to display the data tree. You can then edit variables to correct the data and then try the failed step again. You can also run a service or script and then save changes.

Restriction: You can see data objects only if you have individual or group authority to view or edit them.

Tip: If you notice a stuck process instance, run the BPMReplayOnHoldEMTasks command with the -getNumberOfTasks set to true to see whether there are any on hold Event Manager tasks. Replay the on hold tasks with the -maxNumberOfTasksToReplay or the -bpdInstanceId parameters.

You can choose to skip a step entirely, but use that option with caution. The Skip option passes data to the next activity, but any actions associated with the skipped activity do not run. Therefore, the process might complete, but the result might not be accurate.

Procedure

From the search results, click a failed process instance. A panel opens, displaying information similar to the following:
- Process name, such as Hiring
- Status, which is Failed
- Error message, such as Invalid approval code
- Location, such as Human resources system
- Actions, such as Retry failed steps, Edit data, Escalate, and Skip. For details about all the actions, see Process Inspector actions.
If you select multiple failed process instances and the selections share a single, common error, the common error message is displayed in the process details area. If there are multiple errors that are not common to each process instance, the label Multiple (n) is displayed in the process details area, where n is the number of errors within the selection.

The Expand icon () next to an error message indicates there is additional information available (such as stack trace, location, date-time, and so forth).
To view error details, click the Expand icon ().
From the list that is provided, select one of the following actions:
- To restart the process instance activity, click Retry failed steps and then click OK when a confirmation message is displayed.
- To open the data tree and run a script or service, click Edit data, click in the data tree to edit information, and click Save. You can then click Retry failed steps.