Adovo

Applied AI

A busy LangGraph agent may queue work you meant to reject

Verify multitaskStrategy reject under a busy LangGraph JS stream so duplicate requests are not queued.

November 11, 2026

A user presses Submit again while an agent is still running. The product says “busy; try later,” but a second run starts after the first one finishes. A LangGraph JS change merged on November 10 says StreamOrchestrator.submit() forwarded enqueue on its busy path even when the caller asked for multitaskStrategy: "reject". The fix forwards the requested strategy.

Decide what a second submit means

For each workflow, choose whether a second request should queue, interrupt, roll back or be rejected. A support chatbot may welcome a queue; an agent that changes a customer record may need explicit rejection until the first action is resolved. The strategy is a product contract, not an incidental SDK option.

Test the busy path with a deliberately long first run and a second submit. Verify the actual request sent to the server, the 409 response when rejecting, both error callbacks and the absence of a queued run. The PR notes that the defect was in a specific orchestrator path; do not assume it applies to every LangGraph binding or release. Confirm your installed package includes the merged change.

Keep the user state honest

Disable or label the control while busy, but do not rely on the button alone: two tabs, retries or API clients can still submit. Keep the server strategy authoritative. When a rejection occurs, retain the original run ID and explain that the new request was not scheduled. LangGraph’s JavaScript guidance separates human input, transient errors and recovery paths; the UI should make those states distinct.

Adovo’s view

Concurrency is a customer promise about what happens next. If “reject” can become “later,” the system has changed the user’s intent. Design the promise, test the busy path, and show the outcome plainly.

Continue exploring

Applied AI

LangGraphCaching

A cached LangGraph task may fail on valid mapping keys

Diagnose a TypeError in a cached task before its body runs when mapping keys cannot be ordered.

Read more

Applied AI

LangGraphAgent Reliability

A cancelled LangGraph dev run may still finish its work

Test late cancellation in LangGraph's in-memory dev runtime and bound side effects inside long-running nodes.

Read more

Applied AI

LangGraphStreaming

A hidden Location header can make an agent stream look finished

Test LangGraph JS run-stream reconnection through proxies and CORS before trusting a clean close.

Read more
Explore applied AI services →