Skip to main content
The pipeline canvas is where you assemble graphs: add nodes, connect them with edges, inspect data, and run or debug executions. The editor is built on React Flow, so you get familiar canvas interactions: pan by dragging the background, zoom with trackpad or controls, and snap connections to input and output handles on each node.

​
Drag-and-drop nodes

Open a pipeline and use the node palette (or shortcuts) to place nodes on the canvas. Each node type has a distinct role—sources pull data in, transforms reshape it, destinations write it out, and container nodes group subgraphs for readability.
  • Drag a node from the palette onto empty space, or onto a suggested drop target when the UI offers one.
  • Select a node to open its configuration panel. Settings apply to that node only unless you use shared variables.
  • Move nodes to document your intent: left-to-right flows match how most readers scan a pipeline.

​
Connect with edges

Edges carry datasets from an output handle on one node to an input handle on another.
  1. Click and drag from an output handle (often on the right side of a node).
  2. Release on an input handle on the downstream node.
  3. Planasonix validates that the connection is allowed for those node types and shows errors if the graph would be invalid.
You can have multiple outgoing edges from a node when the graph branches (for example, into a conditional or split). Multiple incoming edges usually mean a join, union, or merge—depending on the downstream node you chose.
Keep edge crossings low and align major stages in columns. Your future self (and reviewers) will thank you during incident response.

​
Preview data at any node

Data preview lets you sample rows at a selected node without running the full production workload.
  • Select a node and open Preview (or the equivalent action in your workspace).
  • Planasonix executes the subgraph needed to produce that node’s output, up to configured limits.
  • Inspect column types, null rates, and sample values to catch schema drift or bad joins early.
Preview uses the same definitions as a full run but typically caps rows and may route work to a smaller compute context depending on your organization’s settings.

​
Run and debug

When you run a pipeline, Planasonix walks the graph in topological order, materializing intermediate results according to your plan and infrastructure.
  • Run from here: Start execution from a selected node (upstream dependencies run as needed).
  • Full run: Execute the entire graph from all root sources through to destinations.
  • Debug: Step through stages, inspect metrics per node, and correlate failures with input row counts or error messages in the run detail view.
Writes to production destinations respect the destination node’s mode (insert, upsert, truncate and load, append). Confirm the target environment and variable values before you run against shared systems.

​
Node types at a glance

Source nodes read data into the graph: databases, APIs, files, CDC streams, webhooks, and similar. They define where data enters and often what slice (table, query, path, or subscription).
See the Nodes overview for the full catalog by category.

​
Right-click context menu

Right-click (or use the context menu shortcut) on the canvas background to add nodes without returning to the palette. Typical actions include:
  • Add node: Pick a type and drop it at the cursor position.
  • Paste (when you copied nodes): Duplicate a subgraph with wiring.
  • Fit view / Zoom: Adjust the viewport when working on large pipelines.
Exact menu entries can vary by plan and enabled features; enterprise workspaces may surface Git or environment actions in adjacent menus.

​
Canvas navigation

Use the canvas toolbar (and keyboard shortcuts where your browser allows) to fit the graph to the viewport, reset zoom, and toggle helpers such as a minimap or grid when your workspace exposes them. On large graphs, collapse or expand container nodes so reviewers focus on one subgraph at a time without losing the overall structure.

​
Next steps

Variables

Externalize hostnames, paths, and flags so one graph works in many contexts.

Source nodes

Configure reads, CDC, webhooks, and iterators.