Analysis & Visualization

Introspect a workflow run's DAG — dependencies, levels, critical path, and stats.

queue.workflows.analyze(runId) returns a WorkflowAnalysis — a read-only view over the run's graph (dag()) and per-node statuses (nodes()). Every method is pure graph computation over that snapshot, so it works whether the run is pending, in flight, or finished.

const a = queue.workflows.analyze(runId);
if (!a) throw new Error("unknown run");

Structure

a.roots(); // ["extract"]            — nodes with no dependencies
a.leaves(); // ["load"]             — nodes nothing depends on
a.ancestors("load"); // upstream    — transitive predecessors
a.descendants("extract"); // downstream
a.topologicalOrder(); // a valid execution order (throws on a cycle)

Levels & critical path

topologicalLevels() groups nodes by dependency depth — everything in one level can run in parallel. criticalPath() returns the longest dependency chain from a root to a leaf (the structural lower bound on how many sequential steps the run takes).

a.topologicalLevels(); // [["extract"], ["transform-a", "transform-b"], ["load"]]
a.criticalPath(); // ["extract", "transform-a", "load"]

Status stats

a.stats();
// { total: 4, byStatus: { completed: 3, running: 1 }, completed: 3, failed: 0, running: 1, pending: 0 }

analyze and wait are separate calls, not chainable: analyze(runId) returns a snapshot now, while handle.wait() resolves when the run finishes. Re-call analyze(runId) after wait() to see the slowest chain on the completed run.