02 · SYSTEM MONITORINGMONITORING

Sentinel

Startup-ready multi-tenant API monitor with automated Celery checks and Slack/Discord alerting.

KEY OUTCOMES
METRIC 01Real-time endpoint down alerts dispatched within 150ms.
METRIC 02Multi-tenant user profiles secured via JWT authentication.
METRIC 03Implicit setInterval frontend polling for zero UI layout shift.
01 · PROBLEM STATEMENT
API downtime or slow responses lead to user churn. Startups need immediate alerts without the high licensing cost of enterprise APM tools.
02 · SYSTEM ARCHITECTURE
Poll StatsPush Monitor JobFetch Poller TaskTrigger Down AlertsRead ConfigurationsWrite Check Run logsReact FrontendFastAPI ServerRedis BrokerCelery WorkerNode AlertsPostgreSQL
03 · DESIGN DECISIONS & TRADE-OFFS
DECISION 01

Celery Task Queue + Redis Broker

Alternative Evaluated:

Asyncio task loops inside FastAPI

Decision Reason:

Guarantees scheduling accuracy and isolates long-running HTTP polling requests outside the main API web thread pool, preventing server blockages.

Trade-off: Adds operational dependency on Redis and Celery processes, managed via container orchestrations.
04 · SYSTEM FAILURE & RECOVERY
Attempted Strategy

Originally built the database schema without proper user boundaries and isolated connections.

Observed Symptom

Database locks and leakage of endpoint alerts occurred between test tenant accounts under heavy simulated traffic.

Resolution & Fix

Secured database queries by implementing strict JWT middleware locks using SQLAlchemy dependencies binding operations directly to authenticated session contexts.

TECHNOLOGY BLUEPRINT
PythonFastAPICeleryRedisNode.jsReact 18PostgreSQL