Skip to main content

Tracing modes

The tracing level defines how much detail is recorded for each node when a workflow runs. You can adapt it to your context: in-depth diagnosis during integration or incident phases, controlled load in day-to-day production.

Deployed version

On an already deployed numbered version, the workflow graph and logic are frozen — only the tracing level remains changeable. See Workflow versions.

Complete​

Each executed node is traced: you get the maximum information to analyze the workflow's behavior (data, metadata, path).

Activation duration in Complete mode​

When you enable the Complete mode, you choose the duration during which it stays applied, between 5 minutes and 4 hours.

When this duration elapses, the level automatically switches back to Performance. This gives you time to investigate while durably limiting the footprint on resources and traces, without having to lower the level manually after the diagnosis.

Standard​

Each executed node is traced, but only failing nodes keep their execution detail in full. Nodes that finish normally are included in the trace without keeping the same level of detail.

This is a useful trade-off when you want to keep an overview of the execution while avoiding storing the detail of every successful step.

Performance​

Only failing nodes are traced. Successful executions do not produce the same per-node detail level.

This mode favors performance and reduced storage usage; it generally suits day-to-day production when you do not need step-by-step diagnosis on every execution.

Choosing the mode on a manual run New​

The form that starts a trigger includes the tracing mode, as on replay. It is prefilled with the workflow mode, or Standard when none is set. Complete shows the performance warning.

Changing the mode when replaying a trace​

When replaying a trace, you can select a tracing mode different from the one of the original execution. This is useful to diagnose an incident in Complete mode without modifying the deployed workflow configuration, then switch back to Performance for the following executions.

Next steps​