Instances & Incidents

After a process is started, the engine creates a process instance that tracks the state of that particular run. This page describes how to view running instances, read incident details, and recover from failures.

Instances and executions

The product uses two labels for the same concept:

  • The top-level navigation tab is called Instances.

  • The tab inside a specific process is called Executions.

Both labels refer to running process instances. See terminology for the full disambiguation.

The instance list

The instance list shows all running process instances for the organization. You can view this list in two places:

  • Instances in the top-level navigation — shows instances across all processes.

  • Executions tab on a specific process — shows instances for that process only.

The list can be filtered by process name, version, status, creation time, and whether the instance has incidents.

Instance detail

Opening an instance shows the process diagram in read-only mode. The diagram highlights the current active task — the step the instance is waiting on.

If any step has an incident, a red badge appears on that element with the incident count. Click the badge to see the incident details.

Incidents

An incident is an error recorded when a process step fails at runtime. Common causes include:

  • A script task that throws an error.

  • A service task that cannot complete — for example, an email that fails to send because the mail server is unreachable.

  • Invalid input parameters for a service task.

Viewing incidents

Incidents can be viewed in three places:

  • Incidents in the top-level Business Processes navigation — a list of all incidents across all processes.

  • Incidents tab on a specific process — incidents for that process only.

  • The incident badge on the instance detail diagram — incidents for a specific step in a specific instance.

Each incident shows the error type, the error message, the affected process and instance, and the time the incident occurred.

Acting on an incident

An incident records the failure. To continue a stuck instance, migrate it to a corrected version or delete it, as described below.

Recovering from a failure

When an instance is stuck due to an incident, you can:

  • Migrate the instance to a corrected process version. If the failure was caused by a problem in the process diagram (for example, a misconfigured service task or a script error), fix the diagram, publish a new version, and migrate the stuck instance to the corrected version. See migrating running instances.

  • Delete the instance. If the instance cannot be recovered, delete it. Deleting an instance removes it and its history permanently.