DevOps and SRE Interview Questions India 2026: Kubernetes, CI/CD and Cloud Guide

Kubernetes is the infrastructure lingua franca at Indian product companies in 2026. Almost every company with 50+ engineers runs Kubernetes for container orchestration. Core Kubernetes concepts tested in interviews: Resource model: Pod (smallest deployable unit), ReplicaSet (maintains N pod replicas), Deployment (manages rolling updates and rollbacks), StatefulSet (for stateful apps like databases, maintains stable network identity and persistent storage), DaemonSet (one pod per node, for monitoring agents and log collectors). Networking: ClusterIP (service accessible only within the cluster), NodePort (exposed on each node's IP), LoadBalancer (creates a cloud load balancer, appropriate for internet-facing services). Ingress (HTTP routing rules that direct traffic to services by hostname or path). Probes: Liveness probe (restart the pod if unhealthy), Readiness probe (only send traffic when the app is ready), Startup probe (for slow-starting apps, delays liveness probe). Resource management: requests (what the pod is guaranteed) vs limits (the maximum the pod can use). OOMKilled: what it means and how to fix it (increase memory limit or find memory leak). CPU throttling: when a pod's CPU limit is hit, the kernel throttles it but does not kill it. Scheduling: node affinity (prefer/require specific node types), pod anti-affinity (spread pods across nodes for HA). Helm: package manager for Kubernetes, manages complex multi-resource deployments as 'charts'. Common troubleshooting questions: 'A pod is stuck in Pending state. What are the possible causes?' (No available nodes, resource limits exceeded on all nodes, image pull error, PersistentVolumeClaim not bound.)

CI/CD pipeline questions test whether you can design release systems that are both fast and safe. CI (Continuous Integration) best practices: every commit triggers a build. The build pipeline: code checkout, dependency install, unit tests (must run in under 5 minutes), integration tests, static analysis (ESLint, Checkstyle), security scanning (Snyk, Semgrep for SAST). The build either passes (merge-ready) or fails with a clear error log. CD (Continuous Deployment/Delivery) strategies: Rolling deployment (replace old pods with new pods gradually, minimum downtime, easy rollback). Blue-green deployment (two identical environments, switch traffic instantly between them, zero downtime, requires 2x infrastructure). Canary deployment (route 5% of traffic to the new version, monitor error rate, then gradually increase to 100%. Best for high-traffic, risk-averse releases like a payment API change). Feature flags: separate deployment from release. Deploy the code, keep the feature off, turn it on for 1% of users, monitor, then roll out. GitOps: ArgoCD and Flux. Application state declared in Git. The cluster polls Git and self-heals toward the declared state. Kubernetes manifests in Git are the single source of truth. Cloud platform questions: AWS is most common in India (used by Flipkart, Swiggy, Zomato, PhonePe). Key services: EC2, EKS (managed Kubernetes), RDS, S3, CloudFront, Lambda, SQS, SNS, API Gateway, IAM. Terraform questions: writing a Terraform module, using remote state in S3, the difference between plan, apply, and destroy. Understanding resource dependencies in a Terraform configuration.

SRE interviews have a strong emphasis on monitoring and incident response because these define the actual day-to-day work. The four golden signals (from Google's SRE book): Latency (how long requests take, including failed requests' latency), Traffic (how much demand is being placed on the system: requests per second), Errors (rate of failed requests, both explicit HTTP 5xx and implicit application-level errors), Saturation (how 'full' the system is: CPU, memory, disk, queue depth). SLI, SLO, SLA definitions: SLI (Service Level Indicator): an actual measurement. 'Our API p99 latency is 180ms.' SLO (Service Level Objective): a target for the SLI. 'We want 99.9% of requests to complete in under 200ms.' SLA (Service Level Agreement): the contractual commitment to customers, with penalties for missing it. Error budget: if your SLO is 99.9% availability (8.77 hours downtime per year allowed), and you have used 5 hours this month, you have 3.77 hours of error budget remaining. Error budgets determine the pace of feature releases vs reliability work. Prometheus and PromQL basics: Counter (always increases: total requests, total errors). Gauge (can go up or down: current CPU usage, queue depth). Histogram (distribution of values: request latency bucketed). PromQL: rate(httprequeststotal[5m]) for request rate. histogramquantile(0.99, rate(httprequestdurationseconds_bucket[5m])) for p99 latency. Incident response process: alert fires, on-call engineer acknowledges (PagerDuty, OpsGenie), triage (what is broken, what is the user impact), mitigation (rollback, restart, scale up, disable a feature), resolution (root cause identified, permanent fix planned), post-mortem (blameless, documents timeline, root cause, and action items to prevent recurrence).

DevOps and SRE is one of India's most under-supplied engineering specialisations in 2026: demand significantly exceeds supply, driving premium compensation. 2026 salary benchmarks: Junior DevOps engineer (0-2 years): Rs 8-22 LPA at product companies. Mid-level (2-5 years): Rs 22-55 LPA. Senior SRE (5-8 years): Rs 50-95 LPA. Principal/Staff SRE: Rs 85-160 LPA at FAANG India. Company-specific benchmarks: Google India SRE is one of the highest-paying SRE roles globally. Atlassian India SRE: Rs 35-120 LPA. Swiggy SRE: Rs 20-75 LPA. PhonePe platform engineering: Rs 22-80 LPA. Certifications that help at the junior-to-mid transition: CKA (Certified Kubernetes Administrator), AWS Solutions Architect Associate, Terraform Associate. These do not substitute for hands-on experience but reduce screening friction. Career trajectory: Junior DevOps (CI/CD focus), Senior SRE (platform engineering ownership, on-call leadership), Principal SRE (reliability engineering strategy, defining SLOs for the organisation), VP of Engineering / Engineering Manager for Platform (people leadership). The SRE role is also a strong path into engineering management: SREs understand the full system and have cross-team relationships that make the EM transition natural.