Provider troubleshooting

Start with the state shown for the specific computer in My computers, then check the local Core Node application. Router cannot repair an offline local runtime from the web dashboard.

The computer is offline

1. Confirm the computer is awake and connected to the internet.
2. Open Core Node and verify the owner account is signed in.
3. Confirm the computer still appears in the local Router flow.
4. Restart only the Core Node provider session if the application shows a stale connection.
5. Return to the dashboard and wait for a fresh observation.

It stays connecting

• Confirm the installed Core Node version supports the current Router protocol.
• Check that the selected model exists locally and finishes loading.
• Look for a model revision mismatch or catalog download request.
• Check local firewall, VPN, or proxy behavior that could interrupt long-lived requests.

It is paused

Verify that sharing is enabled in the supported local flow. A dashboard preference can remain saved while local sharing is stopped. After changing it, wait for observed state to become ready.

It moves between ready and busy

This is normal. A provider accepts one request at a time. While an assignment is active it is busy, then returns to ready after a normal terminal response.

Attempts fail repeatedly

Filter history by provider_failed or interrupted. Compare the error and timing with local runtime logs. Check model memory requirements, local resource pressure, model revision, and whether Core Node was stopped during generation.

Before asking for help

Collect the computer’s displayed state, selected model ID, Core Node version, approximate timestamp and timezone, attempt or request ID, and the exact redacted error. Never include account tokens, API keys, provider secrets, or customer content.

Review readiness and states for the expected lifecycle.