DevOps Engineer Interview Experience for 2-3 YOE | DevOps Interview Questions and Answers 2026

Main ideas / concepts covered (DevOps interview experience)

• The speaker shares a set of DevOps interview questions they faced in a recent interview (for one company).
• They cover key technical topics including:
  • AWS networking and HA/fault tolerance
  • Security and secrets management
  • CI/CD using Jenkins
  • Docker basics (volumes, CMD vs ENTRYPOINT)
  • Kubernetes concepts (deployment strategies and resource types)
  • AWS IAM authentication for DynamoDB from EKS pods (IRSA)
  • Terraform state locking
• They mention having (or providing) a document compiling all answers to the questions.

Detailed methodology / instruction-like sections

AWS: High availability + fault tolerance architecture (initial question)

AWS building blocks mentioned for HA/fault tolerance:

• VPC
• Auto Scaling Groups (ASG)
• ELB / Elastic Load Balancer

AWS: Networking — VPC components question

The speaker explained major VPC components (one-line each):

• Subnets
  • Public subnets
  • Private subnets
• Internet Gateway
• NAT Gateway
• Security Groups
• Network Access Control Lists (NACLs)
• Route tables
• Routes within route tables

AWS Secrets Manager access without internet (VPC endpoint concept)

• Core idea: use VPC endpoints so the service can be accessed without going over the public internet
• Key concept: create a VPC endpoint within the VPC

AWS VPC endpoints: types

Endpoint types mentioned:

• Gateway endpoints
• Interface endpoints

AWS: How to create a VPC endpoint (high-level steps)

A rough console flow (about 4–5 steps):

1. Open VPC endpoints in the AWS console
2. Search for the target service
3. Add/select endpoints configuration
   • The speaker mentions adding “AGs” (unclear term), but the intent is associating the endpoint with the needed networking configuration
4. Follow remaining console steps to create and attach the endpoint

Jenkins CI/CD pipeline for a microservices project (architecture + stages)

Pipeline approach (microservices with similar services)

If the microservices are “almost same”:

• Create one shared Jenkins pipeline
• Use:
  • Build parameters and/or
  • Environment variables
• Optionally: keep one pipeline per service but reuse logic by switching environment variables

Jenkins pipeline stages (deployment workflow)

Stage sequence described:

• Checkout
• Build
• SonarQube analysis
• Push to JFrog (artifact repository)
• Stop the existing service
• Pull the artifact/code (described as pulling the built artifact)
• Stop/backup (speaker mentions a stop and backup step)
• Deploy the new artifact
• Health check
• Decision logic:
  • If health check passes: deployment completes successfully
  • If health check fails:
    • Trigger rollback
    • Run health check again
    • If health check is still not fine: deployment fails and needs manual intervention

Jenkins secrets management

• Emphasized answer: use Jenkins Credential Manager to manage secrets

Docker topics

Docker: Volume types (question + partial recollection)

• The speaker partially recalled Docker volume types.
• Recalled usage included:
  • A “host volume” concept (mounting to a path on the host)
  • A second type (uncertain; possibly bind mount)
• Context: volumes created via docker volume create, then mounted into containers at chosen paths.

Docker: Difference between CMD and ENTRYPOINT

Main difference:

• CMD can be overridden when running docker run
• ENTRYPOINT cannot be overridden directly (default behavior)

Additional note from the speaker:

• With ENTRYPOINT, you can use -- with docker run to change behavior/arguments

Kubernetes topics

Kubernetes: Deployment strategies

Strategies mentioned:

• Blue-green
• Canary
• Rolling update

Kubernetes resources: Deployment vs StatefulSet vs DaemonSet

• StatefulSet
  • Used when you need persistent:
    • Pod identity (persistent pod name)
    • Persistent data
• Deployment
  • For simpler stateless applications
• DaemonSet
  • Ensures pods run on all nodes
• Contrast:
  • In a Deployment, pods can run on any nodes (not necessarily all)

AWS EKS: Access DynamoDB from a pod (IRSA-based)

• Core concept: IRSA (IAM Role for Service Account)

High-level steps described:

1. Create an IAM role
2. Create/configure the trust policy for that role
3. Create a Kubernetes service account (speaker references eksctl)
4. Use eksctl to create/attach service accounts/policies as part of the workflow

Key idea from the speaker:

• If you understand IRSA (IAM role tied to a service account), you can answer this question.

Terraform: State locking and preventing issues

• Concept:
  • If two people run terraform apply simultaneously:
    • Terraform reports that the state is locked
• Options mentioned:
  • Use DynamoDB for remote state locking
  • Or use DynamoDB + S3 for remote state management
• Cross-question note:
  • Without DynamoDB, it may be possible to use S3 locking via a flag/configuration (speaker mentions an S3 flag set to true)

Other topics mentioned (uncertain / partial)

• “What is a dangling image in Docker?”
  • Speaker was not fully aware, but offered an approximate intuition:
    • older/latest-tag images may get removed after rebuilding
  • Details were unclear.
• “Different types of volume” and “dangling image” answers were partially unsure.

Speakers / sources featured

• Speaker: Unnamed individual who performed the interview and narrates their experience.
• Systems/technologies mentioned:
  • AWS (VPC, ELB, NAT Gateway, Internet Gateway, Secrets Manager, VPC Endpoints, DynamoDB, IAM/IRSA, EKS)
  • Jenkins
  • JFrog
  • SonarQube
  • Docker
  • Kubernetes (Deployments/StatefulSets/DaemonSets)
  • Terraform
  • eksctl