Get Your CertAI-300PlayCompete
‹ All AI-300 questions

AI-300 Practice Question 25: Online Endpoint 429

Microsoft Machine Learning Operations Engineer Associate (AI-300) · Test 3 · question 5 of 10

Implement machine learning model lifecycle and operationsmedium

Clients calling a managed online endpoint start receiving HTTP 429 Too many pending requests during busy periods. The deployment's instances are nowhere near CPU saturation, and telemetry confirms the endpoint is well below the managed online endpoint requests-per-second service limit. Which deployment setting should you adjust first?

Select one answer
Show answer and explanation

Answer:

  • A. max_concurrent_requests_per_instance under request_settings, together with the instance count, because the endpoint rejects anything beyond 2 * max_concurrent_requests_per_instance * instance_count requests in flight

Microsoft documents this exact 429: "Your model is currently getting more requests than it can handle. To guarantee smooth operation, Azure Machine Learning permits a maximum of 2 * max_concurrent_requests_per_instance * instance_count requests to process in parallel at any given time. Requests that exceed this maximum are rejected." The fix it points to is to "review your model deployment configuration under the request_settings and scale_settings sections to verify and adjust these settings" (option A). That ceiling is a configured number, which is why it can be hit while CPU is still low.

- B is wrong because request_timeout_ms governs a different failure: when model execution runs longer than that timeout, the endpoint returns 408 Request timeout. Raising it does not change how many requests the deployment will accept in parallel. - C is wrong because it targets the *other* 429 in the documentation, the rate-limited one, raised when requests per second reach the managed online endpoint service limits. The stem rules that out by stating the endpoint is well below that limit. - D is wrong because probe settings are the documented lever for 424 model error cases where liveness or readiness probes fail. Cycling replicas faster does nothing to raise the concurrency ceiling that is rejecting these requests.

Reference: Troubleshoot online endpoint deployment - Azure Machine Learning

Practice Test 3 ›

Study guides for AI-300

AI-200 exam guide: Azure AI Cloud Developer AssociateAI-200 replaced AZ-204: what the Azure AI Cloud Developer exam tests, how it differs from AI-103, price in USD and INR, format and how to prepare.
How AI Is Changing Software Jobs: Where to PrepareHow AI is changing software jobs: what history shows, the evidence on junior developers, what Klarna learned, new AI roles and five domains to prepare for.
Best MLOps certifications 2026: which test real MLOpsThe best MLOps certifications in 2026 from Microsoft, Databricks, Google and AWS, ranked by how much each exam tests MLOps, with cost and order.
Best AI certifications 2026: AWS, Google, Microsoft, ClaudeThe best AI certifications in 2026 from AWS, Google, Microsoft, Claude and Databricks, by level, with price, status and which one to take first.
Microsoft AI certification path 2026: which exam firstMicrosoft AI certification path for 2026 - AI-901, AB-730, AI-103, AI-200, AI-300, AB-100, AI-500 and GH-300 by role, and which exam to take first.
What Is RAG? How Retrieval-Augmented Generation WorksWhat RAG is and how it works: chunking, embeddings, vector search, hybrid search, reranking, GraphRAG and agentic RAG on AWS, Google and Azure.