Live
Dynatrace integrates Arize’s AI observability into its monitoring platformEnabling Node Swap in Kubernetes 1.34: Practical Impact on AI‑Heavy WorkloadsModel Context Protocol trust gaps enable cascading prompt attacksCutting MCP Token Overhead with Codemode: Practical Implications for AI EngineersGitHub secret scanning now detects Lovable Labs, Pydantic, and Supabase credentialsAutonomous code security gains 23‑point boost on CyberGym‑E2E benchmarkGLM 5.3 on Amazon Bedrock: coding‑optimized MoE model with cross‑region inference and prompt cachingAdd SageMaker inference optimization to any coding agent with the aws‑ai‑ml skillDynatrace integrates Arize’s AI observability into its monitoring platformEnabling Node Swap in Kubernetes 1.34: Practical Impact on AI‑Heavy WorkloadsModel Context Protocol trust gaps enable cascading prompt attacksCutting MCP Token Overhead with Codemode: Practical Implications for AI EngineersGitHub secret scanning now detects Lovable Labs, Pydantic, and Supabase credentialsAutonomous code security gains 23‑point boost on CyberGym‑E2E benchmarkGLM 5.3 on Amazon Bedrock: coding‑optimized MoE model with cross‑region inference and prompt cachingAdd SageMaker inference optimization to any coding agent with the aws‑ai‑ml skill
Cloudflare

Cloudflare Durable Objects Dashboard Now Shows Live Traffic Splits

AI SummaryPowered by AI

The Cloudflare dashboard has added a Deployments tab for Durable Object namespaces that displays live versions, traffic percentages, error rates, and wall time metrics directly within the namespace view. This change allows platform engineers to monitor gradual rollout progress without navigating away from their object configuration.

What Changed in the Dashboard

The Cloudflare dashboard has introduced a dedicated Deployments tab for Durable Object namespaces. Previously, verifying the status of a backing Worker deployment required leaving the namespace view to inspect metrics on the main Workers page.

This new interface consolidates visibility by surfacing version information directly alongside specific operational data relevant to objects: requests per second, error rates, and median wall time. The tab is read-only; actions like promoting versions or rolling back changes still occur via the backing Worker's deployment controls.

Understanding Actual vs Configured Traffic

A critical distinction now visible in this view separates configured traffic splits from actual observed shares. When a gradual rollout moves weight to a new version, the dashboard immediately updates the Traffic % column with your configuration settings (e.g., shifting 50/50).

The displayed metrics will lag behind these configurations because individual Durable Objects are pinned to their initial Worker script until you explicitly create a new deployment. Consequently, some objects naturally receive more traffic than others during the transition period.

Operational Implications for Platform Engineers

  • Pinned State Awareness: Remember that observed splits differ from configured ones as long as multiple versions coexist in your namespace due to object pinning logic. Do not assume immediate traffic redistribution upon configuration change.

The dashboard leverages the same GraphQL Analytics API used for standard Workers metrics, meaning data ingestion delays and sampling rules apply equally here. Practitioners should expect a lag of several minutes before actual share percentages fully reflect recent deployment changes in Durable Object analytics compared to general Worker views.

Related CloudNinjas coverage: hands-on guides.

What This Means For Practitioners

This update streamlines the operational workflow for managing stateful logic on Workers. You can now validate that your gradual deployments are progressing as intended without context switching between tabs or services.

To access these metrics, navigate to Workers & Pages, select a namespace under Durable Objects, and click the new Deployments tab. Use this view specifically for monitoring health indicators like error rates during rollout windows rather than making immediate configuration changes from here.

Originally published atCloudflare Developer Platform