Health Monitoring¶
Every AgentCube connector exposes a /health endpoint for monitoring.
Health Endpoint¶
Response Fields¶
| Field | Description |
|---|---|
status | healthy or unhealthy |
version | Connector version (e.g., 2.2.7) |
uptime_seconds | Seconds since container start |
timestamp | Current UTC timestamp |
backend_connected | true if Oracle is reachable (basic mode), null in OIDC mode |
backend_system | Oracle system type (Oracle Essbase or Oracle Planning Cloud) |
backend_url | Configured Oracle endpoint URL |
connector_id | Connector identifier (agentcube-essbase or agentcube-planning) |
sdk_version | AgentCube SDK version embedded in this build |
auth_mode | Current authentication mode (basic or oidc) |
oidc_issuer | OIDC discovery URL (OIDC mode only, null in basic mode) |
backend_note | Additional context (e.g., OIDC mode backend verification note) |
MCP Connector Status Tool¶
The mcp_connector_status tool returns the same information as the health endpoint but is callable from the AI platform. This is useful for verifying connectivity without leaving the AI conversation.
Monitoring Recommendations¶
- Poll
/healthperiodically — set up a health check in your monitoring system (e.g., Azure Monitor, Datadog, Uptime Robot) - Alert on
status: unhealthy— indicates the connector cannot reach Oracle - Monitor
uptime_seconds— unexpected resets indicate container restarts - Check after deployments — verify
versionmatches the deployed version
OIDC Mode Considerations¶
In OIDC mode, backend_connected is always null. This is expected — backend connectivity cannot be verified without a user context. The backend is verified on each authenticated user request.
To verify backend connectivity in OIDC mode:
- Authenticate via your AI platform
- Run
test_datasource— this makes an authenticated request to Oracle and confirms connectivity