WEB API process model restart and error handling

In Appian, you cannot restart a failed process instance or resume from a specific node once it has errored or completed.
Instead, the recommended approach is to design retry logic—store the required data and trigger a new process or subprocess to re-run only the failed steps.
To make this safe, ensure your integrations are idempotent and state is tracked, so retries don’t cause duplicate or inconsistent results.

No you can't. And this is no technical limitation, but a logical. Something that break in the middle of multiple steps is in an unknown state. You will either have to make the support team fix that process manually, or design a error handling process.

Appian does not directly provide an option to restart a failed process instance from the exact broken node. The better approach is to design exception handling with retry logic or create a separate recovery process that uses the Process ID to re-run the failed asynchronous subprocess or integration. In short, this is usually managed through process design rather than manual restart from the failed step.

No. Does not support programmatically resuming a failed process instance from the broken node.
Only option natively: Manual retry via Process Monitor.

Practical workaround: Store async subprocess inputs in a DB -> on failure, expose a retry endpoint that re-invokes only the failed subprocess using stored inputs.