Why
Root-cause categories—memory pressure, lock contention, config drift, missing backpressure—not blame. Tie symptom to mechanism (e.g. heap exhaustion vs leak vs oversized payload).
Production systems
Real outages rarely start with “fix line 42.” They start with symptoms—OOM, stuck threads, stale cache, duplicate charges. Each guide in this track answers the same three questions: Why does this happen in production? What do you check first (in order)? How do you fix it now and prevent it next time?
The Why · What · How framework
What
Ordered checklist: metrics → logs → thread dump → heap dump → recent deploy → dependency health. What to capture before restart destroys evidence.
How
Immediate mitigation (scale, circuit break, kill feature flag), durable fix (code, pool sizing, idempotency), and guardrails (alerts, load tests, runbooks).
JVM & runtime health
Memory, garbage collection, and processes that degrade over time.
-
Guide
Your application suddenly throws
OutOfMemoryError. How will you debug it? -
Guide
You suspect a memory leak. How will you confirm it and find the retaining path?
-
Guide
Frequent GC pauses are impacting latency. How will you tune and verify improvement?
-
Guide
The application becomes slow after a few hours of uptime. What will you check first?
- Guide
- Guide
-
Guide
Metaspaceor class-loader leaks appear after many redeploys. How do you diagnose them?
CPU, threads & concurrency
High CPU with low traffic, blocked threads, deadlocks, and unsafe shared state.
-
Guide
CPU usage is ~90% but request traffic is very low. What could be wrong?
-
Guide
A thread is stuck in
BLOCKEDstate. How will you identify what it is waiting on? -
Guide
A deadlock is detected in production. How will you resolve it and prevent recurrence?
-
Guide
The request thread pool (or
ForkJoinPool) is exhausted under load. What is your approach? -
Guide
You need to handle very high concurrency safely. What design patterns and primitives will you use?
-
Guide
The service becomes randomly unresponsive for short windows. What could be happening internally?
-
Guide
HashMap(orConcurrentHashMap) performance drops under heavy load. Why?
Production vs local & observability
Works on my machine, inconsistent logs, and tracing one request end to end.
-
Guide
The API works locally but fails in production. What will you investigate first?
-
Guide
Logs show inconsistent behavior across requests (same input, different outcome). Why?
- Guide
-
Guide
Support cannot correlate user complaints to logs across services. What do you add to the platform?
- Guide
-
Guide
The gateway returns 502/504 but Kubernetes reports pods healthy. Where do you look next?
Database & persistence
Slow queries, pool exhaustion, and data-layer bottlenecks.
-
Guide
Database calls are slowing down the service. What optimizations will you apply?
-
Guide
The DB connection pool is exhausted and requests hang. How do you fix and size pools correctly?
-
Guide
One API endpoint triggers hundreds of SQL queries per request. How do you find and fix N+1?
-
Guide
Spikes in DB row lock wait time appear during peak traffic. What changed and what mitigations apply?
-
Guide
Users report stale reads after an update. How do you handle read-replica lag in the application?
Caching & consistency
Stale data, stampedes, and cache-aside pitfalls.
Scaling, resilience & API correctness
Horizontal scale that does not help, duplicates, and backpressure.
-
Guide
You added more instances but latency did not improve. Why might horizontal scaling fail?
-
Guide
The system processes duplicate requests (retries, double-clicks). How will you make APIs idempotent?
- Guide
- Guide
-
Guide
Kafka consumer lag grows for hours after a traffic spike. What is your triage order?
- Guide