Reaktor Trace Walk
The Diagram tab on a Reaktor Deployment draws the Aktor graph and lets you step through a recorded run node by node. You see the execution order, what each Aktor received and returned, how long it took, and where a run failed. It is the visual counterpart to the Logs tab: same run data, shown on the diagram instead of as text.
Open the diagram
Open a Reaktor Deployment and select the Diagram tab. The diagram shows every Aktor, database, and API in the Reaktor and how they connect. The Reaktor diagram is generated from your code, so it always matches the deployed version.
Three modes
A mode switch at the top of the diagram controls what the diagram shows.
| Mode | What it shows |
|---|---|
| Static | The structure only. No run data, no step controls. Use it to read the graph. |
| Trace | A recorded, finished run that you step through manually or play back. |
| Live | The current run, updated as it executes, with the diagram advancing on its own. |
Trace and Live need a captured trace. Switch Debug on in the toolbar, then run the Reaktor once. With Debug off, a run records no trace and the diagram reports that none exists. See Run Tracing for what a trace stores.
Step through a recorded run
Switch to Trace mode and pick a run from the run selector, which lists the most recent runs. Then move through the steps:
- Start / Previous / Next / End: jump to the first step, move one step, or jump to the last step.
- Position: type a step number to go straight to it.
- Play / Pause: advance through the steps on a timer. The pace adapts to each step's duration.
- Value changes: stop only on steps where a node's value changed, skipping steps with no visible effect.
As you move, the diagram colors each node by its state and marks the current step.
Follow a running Reaktor
Switch to Live mode to watch a run as it happens. The diagram follows the active execution and advances to each new step as the platform records it. A badge shows which run you are following. Use this to watch a long-running or scheduled Reaktor without reloading.
Inspect a node
Click a node to pin it. The detail pane on the right shows:
- Header: the node label, its type, and its state in the current step.
- Identity and timing: the node ID (copyable), the execution time in milliseconds, and the step number.
- Error: the error message, when the node failed.
- Output: the value the Aktor returned, as JSON.
- Inputs: each incoming edge, the upstream node it came from, the parameter name, the relation (child or reference), and that input's value.
The inputs section is the fastest way to trace a wrong value back to its source: follow the edges upstream until the value first goes wrong.
Focus the view
Large Reaktors produce large diagrams. Three controls keep them readable:
- Collapse level: collapse or expand groups of nodes. Start fully collapsed and open only the branch you care about.
- States: color nodes by run state and show a legend (ready, running, done, error, current).
- Filter: hide node types you do not need, such as constants or inputs, and toggle reference edges.
The collapse and filter settings are local to this view and persist while you step or while a Live run updates.
Trace Walk vs the Logs tab
Both surfaces read the same recorded run, from different angles.
- Reaktor Trace Walk (this page, Diagram tab): the run laid over the graph. Best for understanding flow, order, and where a branch went.
- Reaktor-Trace tree (Logs tab): the run as a step-by-step list. Best for scanning many steps quickly and reading raw inputs and outputs.
Use the diagram to locate the step, then the tree (or the Logs tab) for the full text.
Common mistakes
- Expecting a trace without Debug Mode. Static mode always works, but Trace and Live stay empty until Debug Mode is on and the Reaktor has run.
- Reading old runs in Live mode. Live follows the current execution only. To inspect a finished run, switch to Trace and select it.
- Losing nodes behind collapsed groups. A node you expect may sit inside a collapsed group. Raise the collapse level or open the group before concluding a step did not run.
- Confusing reference edges with data flow. Reference inputs are marked separately from child inputs. A reference does not mean the value flowed along that edge in this run.
For handling failures in code so they surface cleanly here, see Error Handling.