Anti-Vibe Networking Audit

- [ ] Deadlocks
- [ ] Race conditions
- [ ] Distributed locks
- [ ] Lock timeouts
- [ ] Lock cleanup after crashes
- [ ] Idempotent operations
- [ ] Duplicate request handling

- [ ] N+1 queries
- [ ] Slow queries
- [ ] Missing indexes
- [ ] Connection pool exhaustion
- [ ] Transaction boundaries
- [ ] Long-running transactions
- [ ] Read/write contention

- [ ] Eventual consistency behavior defined
- [ ] Stale reads handled
- [ ] Write conflicts handled
- [ ] Ordering assumptions documented
- [ ] Duplicate events handled
- [ ] Out-of-order events handled
- [ ] Retry-safe writes

- [ ] Failover tested
- [ ] Partial failure handled
- [ ] Dependency outage behavior
- [ ] Timeout behavior
- [ ] Retry strategy
- [ ] Exponential backoff
- [ ] Retry jitter
- [ ] Circuit breakers
- [ ] Graceful degradation

- [ ] Load balancing
- [ ] Uneven node distribution
- [ ] Hot partitions
- [ ] Rate limiting
- [ ] Backpressure
- [ ] Queue saturation
- [ ] Connection limits
- [ ] Thundering herd prevention
- [ ] Traffic spikes tested

- [ ] Memory leaks
- [ ] Socket leaks
- [ ] Connection leaks
- [ ] File descriptor exhaustion
- [ ] Unbounded queues
- [ ] Unbounded caches
- [ ] Worker/thread exhaustion
- [ ] Cleanup on disconnect

- [ ] Connection timeouts
- [ ] Read timeouts
- [ ] Write timeouts
- [ ] Keep-alive behavior
- [ ] Connection reuse
- [ ] DNS failure handling
- [ ] TLS failures
- [ ] Packet loss tolerance
- [ ] High-latency behavior
- [ ] Network partitions

- [ ] Request size limits
- [ ] Response size limits
- [ ] Version compatibility
- [ ] Schema compatibility
- [ ] Breaking-change protection
- [ ] Idempotency keys
- [ ] Pagination
- [ ] Partial-response handling
- [ ] Dependency fan-out controlled

- [ ] Dead-letter queue
- [ ] Poison message handling
- [ ] Consumer lag monitoring
- [ ] Message acknowledgement behavior
- [ ] At-least-once delivery accounted for
- [ ] Replay behavior tested
- [ ] Queue backlog recovery
- [ ] Producer/consumer failure scenarios

- [ ] Rollback strategy
- [ ] Database rollback strategy
- [ ] Forward-compatible migrations
- [ ] Blue/green or canary safety
- [ ] Mixed-version compatibility
- [ ] Restart recovery
- [ ] Disaster recovery tested
- [ ] Backup restore tested
- [ ] RPO/RTO defined

- [ ] Request tracing
- [ ] Correlation IDs
- [ ] Error rates monitored
- [ ] Latency percentiles monitored
- [ ] Queue depth monitored
- [ ] Connection pool usage monitored
- [ ] Retry rates monitored
- [ ] Failover events visible
- [ ] Alerts are actionable

- [ ] TLS everywhere appropriate
- [ ] Secrets not transmitted in logs
- [ ] Auth between services
- [ ] Authorization enforced server-side
- [ ] Internal endpoints protected
- [ ] Replay attacks considered
- [ ] Rate-limit abuse considered
- [ ] Input validation at trust boundaries

- [ ] Kill one service mid-request
- [ ] Kill the database connection
- [ ] Add 500–2000 ms latency
- [ ] Drop random requests
- [ ] Duplicate messages
- [ ] Deliver events out of order
- [ ] Restart nodes under load
- [ ] Exhaust the connection pool
- [ ] Simulate one unhealthy region/node
- [ ] Verify the system recovers without manual cleanup