Troubleshooting
Troubleshooting
Use this page when IO Gateway starts but requests, dashboard actions, provider accounts, or release checks do not behave as expected.
Diagnostic order
- 1Check process health.
Start with
/health,/ready, and service logs. - 2Check authentication.
Separate dashboard auth failures from client API-key failures.
- 3Check provider accounts.
Look for disabled, expired, cooling-down, failed, or quota-exhausted accounts.
- 4Check routing rules.
Inspect custom models, managed-key scopes, priority routing, and fallback behavior.
Common symptoms
| Symptom | Likely check |
|---|---|
401 from client request | Wrong shared proxy key or managed API key. |
| Dashboard login fails | Admin auth key, TOTP secret, cookie security, or session settings. |
| Model alias not found | Custom model name, ctm: prefix, or saved alias state. |
| Unexpected account used | Priority routing, account health, key scope, or custom-model fallback. |
| Notification missing | Destination token/webhook, outbound network access, or dashboard notification settings. |
Useful commands
curl http://127.0.0.1:8319/health
curl http://127.0.0.1:8319/ready
curl http://127.0.0.1:8319/api-docs/openapi.json