50 senior-level, scenario-based questions across 8 areas
Showing 50 of 50 questions
Design a CI/CD pipeline for 50 microservices where each service can be independently built, deployed, and rolled back.
A bug appears only in production, never in staging, even though 'the same code' was deployed. How do you prevent this class of problem?
Design a Git branching strategy for 50 engineers shipping daily. What prevents merge conflicts and broken mains?
Design a safe release strategy for a critical API that must never take downtime and must limit the blast radius of a bad deploy.
Your Kubernetes deployment completed successfully, but users are receiving 503 errors. Walk through your debugging approach.
Pods are stuck in CrashLoopBackOff after a deploy. How do you find and fix the cause?
CPU usage and latency are climbing, but the Horizontal Pod Autoscaler is not adding pods. Why, and how do you fix it?
Your service image is 1.2 GB and scale-ups are slow because nodes take too long to pull it. How do you reduce image size and speed up starts?
Design a zero-downtime deployment strategy for a critical API on Azure.
An automated Azure deployment fails with an authorization error when creating resources. How do you diagnose and fix it without weakening security?
Kafka consumer lag is continuously increasing while producer traffic remains constant. How would you investigate and fix it?
API latency jumped 10x after a deploy. Using metrics, traces, and logs, how do you localise the cause?
Secrets are currently hardcoded in pipeline scripts and committed env files. How do you manage secrets across CI/CD safely?
Flaky integration tests randomly fail the pipeline and engineers have started re-running until green. How do you fix this without disabling testing?
Your org is deciding between a monorepo and many repos for 50 services. How does this choice affect CI/CD, and what do you recommend?
How do you keep Jenkins pipelines maintainable across dozens of services instead of copy-pasted job configs?
Builds are queuing for 20+ minutes because Jenkins agents are saturated. How do you scale and speed up the build farm?
You want blue-green deploys, but the release includes a database schema change. How do you avoid breaking the old version during cutover?
Design an automated rollback strategy so a bad release recovers without waiting for a human to notice.
How do feature flags change your deployment strategy, and what risks do they introduce?
A kubectl rollout is stuck and never completes, but the old pods still serve traffic. What is happening and how do you resolve it?
Explain how CPU and memory limits behave differently in Kubernetes, and how to set requests and limits correctly.
How should application configuration and secrets be delivered to pods, and what changes when a value updates?
Explain ClusterIP, NodePort, and LoadBalancer Services, and how in-cluster DNS resolves a Service.
How do you harden a container image and its runtime for production?
Pods are being evicted and rescheduled seemingly at random. What causes eviction and how do you prevent it?
How do App Service deployment slots enable zero-downtime releases, and what are the gotchas?
How do you store and access application secrets in Azure without putting them in config files or code?
When would you choose App Service, Azure Functions, or AKS for a workload?
Why manage Azure infrastructure as code with Bicep/ARM, and what makes an IaC deployment safe to re-run?
Design autoscaling for an Azure workload with spiky traffic. What signals and guardrails do you use?
How do you keep an App Service or AKS workload's traffic to databases and storage private?
Cloud spend is climbing faster than traffic. How do you find and control Azure cost without hurting reliability?
A payment API sometimes charges customers twice when clients retry after a timeout. How do you make write APIs safe to retry?
How do you protect an API from abuse and overload with rate limiting, and where do you enforce it?
How do you evolve a public API without breaking existing clients?
You need reliable background processing for long-running jobs off the request path. How do WebJobs/Functions fit, and how do you make them reliable?
Design authentication and authorization for a set of microservice APIs.
A downstream system is receiving duplicate Kafka messages. How do you reason about delivery semantics and achieve effectively-once processing?
You must process events for a given user strictly in order, but also need high throughput. How do partitions and keys help?
Your consumer group keeps rebalancing and pausing consumption. What causes rebalancing storms and how do you stop them?
A poison message keeps crashing your consumer and blocking the partition. How do you handle retries and bad messages?
Explain metrics, logs, and traces — when each is the right tool, and their cost trade-offs.
Your team is drowning in noisy alerts and starting to ignore them. How do you design alerting around SLOs and error budgets?
A request crosses six services and you cannot tell where time is spent. How does distributed tracing solve this and what does it require?
A service's memory grows steadily until it OOMs every few days, then restarts. How do you find and fix it?
One slow downstream dependency is dragging your whole service down. How do you stop the cascade and prevent it recurring?
Under load, requests start failing with 'no available connections' from the database. What is happening and how do you fix it?
When a hot cache key expires, a flood of requests hits the database simultaneously and it falls over. How do you prevent this?
Users report occasional timeouts you cannot reproduce. How do you systematically track down an intermittent failure?