Run and debug

Open a workflow run record to inspect its status, node inputs and outputs, and logs. After processing completes, also check the results in the target knowledge base, table, or volume.

Inspect a run

  1. Select the run record and confirm the workflow version and input data.

  2. Find the node that failed or produced unexpected output.

  3. Check its input fields, file references, and parameters.

  4. Read its output and logs to locate the error.

  5. For incorrect inputs, check upstream outputs and bindings. For incorrect settings, update the node and run again.

Common problems

Symptom

What to check

A node keeps waiting

Inspect the run status and errors; if it continues waiting, provide the run information to an administrator

An input is missing or empty

Check that the upstream output contains the required field and that the downstream binding uses the same name

Results differ from the canvas configuration

Confirm the executed version, node order, and input bindings

A sub-workflow cannot be called

Confirm that the child flow is defined in the current workflow; use Workflow Trigger to start another published workflow

A text-processing node receives no text

Check whether a file ID was passed; parse the file first, then bind its text or document output

SQL execution fails

Check syntax, target tables, and referenced fields; see SQL Processing

A node completes but the result is unexpected

Check output counts, content, and destination; for external requests, also inspect the HTTP status code and response body

Run again after making changes

Validate changes with a small sample and inspect key outputs step by step. Before repeating writes to tables, indexes, or files, confirm the destination and overwrite settings to avoid duplicates or overwriting results you need to retain.

See Variables and data flow for input and state references.

Last updated on