Skip to main content

Monitor and Debug

How to find what went wrong with a workflow execution, inspect variable values, and understand the execution timeline.

Finding Executions​

Navigate to Run in the top navigation bar (or go to /process-instances).

  1. Select your workflow from the PROCESS dropdown on the left panel
  2. The instance list shows all executions for that workflow, newest first
  3. Each item shows: instance ID (truncated), status, start time, and duration

Instance statuses:

StatusMeaning
Active (green)Currently running
Waiting (orange)Paused at an Input node or waiting for an event
Failed (red)Execution encountered an error and stopped
Completed (blue)Finished successfully

Click any instance to open its detail view on the right.

Reading the Overview Tab​

The Overview tab (default) shows the high-level execution metadata:

  • Status — current state
  • Version — which version of the workflow ran
  • Started — when execution began
  • Duration — how long it ran (or how long it has been running)
  • Instance ID — the full UUID (copyable) — useful for support requests or API queries
  • Process ID — the workflow's UUID

Inspecting Nodes​

Click the Nodes tab to see every node that executed, in order.

Each node shows:

  • Node number and name (e.g., "#1 Start Event", "#2 AI Agent")
  • Status badge (see node statuses below)
  • Node type
  • Start timestamp
  • Duration

Node statuses:

StatusMeaning
CompletedNode finished successfully
ErrorNode failed — check for error details in variables
In ProgressNode is currently executing
CancelledNode was cancelled before completing

Look for the node with Error status to find where the execution failed.

Inspecting Variables​

Click the Variables tab to see all process variables at the current state of the execution.

This is the most useful debugging tool:

  • If an AI node failed, check what {prompt} or {userInput} variables contained
  • If a REST API node failed, check the input variables that were passed to the request
  • If an If/Else condition routed incorrectly, check what value the condition variable held

Variables are shown as key-value pairs. JSON values are pretty-printed.

Following the Execution Timeline​

Click the Timeline tab to see the execution in chronological order:

  • Each node appears as a card with its start time and duration
  • Use this to understand the sequence of events and spot which stage took the most time

Common Failure Patterns​

AI Node Failed​

  1. Go to the Variables tab — check what the AI received as input
  2. Look for variables with null or unexpected values — these indicate missing upstream data
  3. Check that your AI provider connection is still valid (Go to Connect → find the connection → verify credentials)
  4. Verify the prompt uses {varName} (curly braces only, no dollar sign)

REST API Node Failed​

  1. Check the Variables tab for serviceTaskStatusCode — a 401 or 403 means authentication failed
  2. Check that your connection credentials are correct
  3. Check the URL and path variables — a 404 means the endpoint doesn't exist
  4. Look for null variables in your URL path — {orderId} not resolving means the variable was never set

Flow Not Triggering​

If a webhook or scheduled trigger is not starting the workflow:

  • Verify the workflow is published (DRAFT flows do not fire on triggers)
  • Check Administration → Webhooks — confirm the endpoint is active
  • For webhooks: verify the external service is sending to the correct URL
  • Check the signature secret matches what the external service is using

Execution Is Stuck (Waiting Status)​

Executions in Waiting status are paused at an Input node waiting for user input. This is expected behavior when a human task is pending.

To see which node it is waiting on, check the Nodes tab for a node in "In Progress" state.

To stop a stuck execution, click Abort in the instance detail header (only visible for Active/Waiting executions, requires abort permissions).

Aborting an Execution​

Click Abort (visible in the instance detail header for Active or Waiting executions) to cancel the execution. A confirmation dialog appears before aborting.

Aborted executions show Failed status in the list.

Admin Analytics​

For aggregate metrics across all executions, see Administration → Observability:

  • Process Analytics — execution counts, success rates, error hotspots, slowest nodes
  • AI Analytics — token usage, cost, model performance
  • Dead Letter Queue — failed node executions that can be retried or discarded

See Process Analytics for details on each analytics page.