Health Check API: Monitor Sentinel Service Dependencies

Sentinel exposes two health endpoints: a lightweight liveness probe at GET /health (root level) and a full integration health check at GET /api/health. The integration check pings all six external service dependencies with a 3-second timeout and returns an aggregate status that reflects the overall health of the system. Both endpoints require no authentication, making them suitable for use with uptime monitors, container health checks, and load balancer probes.

GET /health

Simple liveness probe. Returns immediately with a static healthy response. Use this endpoint for container HEALTHCHECK directives and load balancer health probes. Auth required: No

curl https://sentinel-softserve.onrender.com/health

{"status": "healthy"}

Always returns	`200 OK`
Response body	`{"status": "healthy"}`

GET /api/health

Full integration health check. Pings all six service dependencies sequentially, measures latency, and returns an aggregate status. Use this endpoint for dashboards, alerting, and diagnosing connectivity issues. Auth required: No

Services Checked

Service key	Environment variable	Default URL
`prometheus`	`PROMETHEUS_URL`	`http://localhost:9090`
`loki`	`LOKI_URL`	`http://localhost:3100`
`chromadb`	`CHROMA_HOST`	`http://localhost:8001`
`alertmanager`	`ALERTMANAGER_URL`	`http://localhost:9093`
`langfuse`	`LANGFUSE_HOST`	`http://localhost:3001`
`supabase`	`SUPABASE_URL` (via client)	Tested with `SELECT id FROM incidents LIMIT 1`

HTTP services are considered reachable if any response is received — even 401 Unauthorized or 404 Not Found. Only connection errors and timeouts are treated as failures. The Supabase check uses the Supabase PostgREST client directly rather than an HTTP ping.

Aggregate Status Logic

The top-level status field in the response is derived from the number of failed service checks:

Error count	Aggregate status
0	`healthy`
1–2	`degraded`
3 or more	`critical`

Response Schema

status

string

Aggregate health status: healthy, degraded, or critical.

services

object

Map of service names to their individual health results.

services.[name].status

string

ok if reachable, error if the connection failed or timed out.

services.[name].url

string

The URL that was pinged. Omitted for supabase.

services.[name].latency_ms

integer

Round-trip latency in milliseconds. Present only when status is ok. Omitted for supabase.

services.[name].error

string

Error description. Present only when status is error. One of "Connection refused", "Timeout (3s)", or a generic exception message.

Example Responses

All healthy
Degraded (1 error)
Critical (3+ errors)

{
  "status": "healthy",
  "services": {
    "prometheus": {
      "status": "ok",
      "url": "http://localhost:9090",
      "latency_ms": 12
    },
    "loki": {
      "status": "ok",
      "url": "http://localhost:3100",
      "latency_ms": 8
    },
    "chromadb": {
      "status": "ok",
      "url": "http://localhost:8001",
      "latency_ms": 15
    },
    "alertmanager": {
      "status": "ok",
      "url": "http://localhost:9093",
      "latency_ms": 6
    },
    "langfuse": {
      "status": "ok",
      "url": "http://localhost:3001",
      "latency_ms": 23
    },
    "supabase": {
      "status": "ok"
    }
  }
}

{
  "status": "degraded",
  "services": {
    "prometheus": {
      "status": "ok",
      "url": "http://localhost:9090",
      "latency_ms": 11
    },
    "loki": {
      "status": "error",
      "url": "http://localhost:3100",
      "error": "Connection refused"
    },
    "chromadb": {
      "status": "ok",
      "url": "http://localhost:8001",
      "latency_ms": 14
    },
    "alertmanager": {
      "status": "ok",
      "url": "http://localhost:9093",
      "latency_ms": 7
    },
    "langfuse": {
      "status": "ok",
      "url": "http://localhost:3001",
      "latency_ms": 19
    },
    "supabase": {
      "status": "ok"
    }
  }
}

{
  "status": "critical",
  "services": {
    "prometheus": {
      "status": "error",
      "url": "http://localhost:9090",
      "error": "Connection refused"
    },
    "loki": {
      "status": "error",
      "url": "http://localhost:3100",
      "error": "Connection refused"
    },
    "chromadb": {
      "status": "error",
      "url": "http://localhost:8001",
      "error": "Timeout (3s)"
    },
    "alertmanager": {
      "status": "ok",
      "url": "http://localhost:9093",
      "latency_ms": 5
    },
    "langfuse": {
      "status": "ok",
      "url": "http://localhost:3001",
      "latency_ms": 21
    },
    "supabase": {
      "status": "error",
      "error": "could not connect to server"
    }
  }
}

Usage Examples

cURL
Docker HEALTHCHECK
Python (requests)

curl https://sentinel-softserve.onrender.com/api/health | jq .status

HEALTHCHECK --interval=30s --timeout=5s --start-period=10s --retries=3 \
  CMD curl -f http://localhost:8000/health || exit 1

import requests

response = requests.get("https://sentinel-softserve.onrender.com/api/health", timeout=10)
health = response.json()

if health["status"] == "critical":
    send_pager_alert(f"Sentinel is CRITICAL: {health['services']}")

For production uptime monitoring, poll GET /health (liveness) every 30 seconds and GET /api/health (integration) every 60 seconds. Alert on critical status; notify (but do not page) on degraded.

Configuring Service URLs

Override the default service URLs via environment variables on the backend container:

PROMETHEUS_URL=http://prometheus:9090
LOKI_URL=http://loki:3100
CHROMA_HOST=http://chromadb:8001
ALERTMANAGER_URL=http://alertmanager:9093
LANGFUSE_HOST=http://langfuse:3001

In Docker Compose deployments, use the service names as hostnames (e.g. http://prometheus:9090) rather than localhost.

Overview

Endpoints

Health Check API: Monitor Sentinel Service Dependencies

GET /health

GET /api/health

Services Checked

Aggregate Status Logic

Response Schema

Example Responses

Usage Examples

Configuring Service URLs

Build docs developers (and LLMs) love

Overview

Endpoints

Documentation Index

​GET /health

​GET /api/health

​Services Checked

​Aggregate Status Logic

​Response Schema

​Example Responses

​Usage Examples

​Configuring Service URLs

Build docs developers (and LLMs) love

GET /health

GET /api/health

Services Checked

Aggregate Status Logic

Response Schema

Example Responses

Usage Examples

Configuring Service URLs