Debug a workflow
Test run
From the editor header, start a test run. It executes the workflow in its current editor state — no need to save first — and streams logs into the panel at the bottom.
Constraints, all deliberate:
one test at a time per workflow (
CONCURRENT_TEST_IS_FORBIDDENotherwise),the run survives a page reload and is resumed when you return,
leaving the editor asks for confirmation, then terminates the run,
at least one step is required.
Read the log tree
The tree mirrors the workflow. Expand a step to get its call:
HTTP method, endpoint URL, status code and execution time,
request and response headers,
request and response body.
The copy icon puts a header, body or URL on the clipboard.
Loop iterations are grouped, with a pager (2 / 12) that also jumps to a
specific index. Entries are marked ERROR, WARNING or INFO.
When a run fails the viewer takes you to the failure rather than making you hunt: red trace dots, nested loops paged to the failing iteration, and the error shown on the step instead of in the request/response section.
Panel controls: fullscreen, clear, minimise.
Historical runs
Scheduled executions use the same viewer. In the schedule list, Last Success and Last Fail carry a see logs link. It opens the Execution Logs dialog, which first asks you to pick an execution from a drop-down — the retained runs of that schedule — and then renders its tree.
Note
Logs only exist for schedules with logging enabled (Debug column).
Retention defaults to the 2 most recent successful and 3 most recent failed runs
per workflow, configurable under log.retention.per-connection — see
Configuration.
Errors before execution
Some problems are caught at save or test-start and reported against the node, which is outlined in red:
Message |
Cause |
|---|---|
|
An |
|
A step references a deleted connector. |
|
A test or schedule for this workflow is already running. |
Field type mismatches during binding are reported as readable messages rather than a stack trace.
Support bundles
When you need support to look at a run, generate a support bundle from the schedule’s Support logs action. It runs the workflow, collects the logs, and stores them with the invoker files as a ZIP.
Because logs contain real payloads, you choose what leaves your system:
Level |
URL |
Headers |
Request |
Response |
|---|---|---|---|---|
Light |
masked |
visible |
visible |
visible |
Medium |
masked |
masked |
visible |
visible |
Strict |
masked |
masked |
masked |
visible |
Custom |
free |
free |
free |
free |
Or toggle each section individually with its eye icon. You are notified when collection starts and again — over the WebSocket — when the file is ready.
Bundles are listed under Configurations → Support Files, where a green status means a success archive exists and red means only a failure archive was produced.
If nothing streams
An empty log panel and a dashboard reporting Connection lost both point at the
same thing: /ws is not proxied correctly. See
Troubleshooting.