Accenture Interview Round 1 & 2 —

DevOps Technical Deep Dive

LinkedIn: Amit Singh

Medium: Amit Singh – Medium

Q. How are your day-to-day activities as a DevOps Engineer?

Pointers

Monitoring infrastructure
Writing/maintaining IaC
Improving pipelines
Handling releases
Supporting dev teams
Troubleshooting incidents

Scenario style

“On a typical day, I monitor production clusters, manage and improve our Jenkins pipelines,
work on Terraform modules for resource provisioning, and collaborate with developers for
application releases. I also handle incidents like pod failures or networking issues.”

🧠 Easy to remember: Monitor, Improve, Support, Troubleshoot

Q. What is a NAT Gateway?

Pointers

Allows private subnets to access the internet

Without exposing resources to incoming traffic
Managed by AWS
✅ Scenarios
“We had EC2 instances in private subnets that needed to pull Docker images from public
repos. We placed a NAT Gateway in a public subnet, routing private subnet traffic through it
so those instances could reach the internet safely.”

“In my last project, the private subnets for EC2 instances needed to pull OS updates from the
internet. I used a NAT gateway in the public subnet, routing traffic through it, while blocking
inbound access to those instances.”

Diagram

Internet-> IGW → NAT Gateway → Private Subnet EC2

Advanced Q&A
Q: Can a NAT Gateway receive inbound traffic?
A: No — it only handles outbound requests from private subnets.

Best practices

Always place a NAT gateway in a public subnet

Remember, private subnet routes point to the NAT

Q. Prerequisites to upgrade a Kubernetes cluster

Pointers

Backup etcd
Drain nodes
Check deprecated APIs
Update kubectl
Test in staging

✅ Scenario style
“We planned a Kubernetes upgrade from v1.24 to v1.27. First, we backed up etcd, verified
compatibility of Ingress controllers, and checked deprecated APIs. Then we drained and
upgraded worker nodes in a rolling manner.”
🧠 Easy to remember: Backup, Drain, Check, Test

Q. What is Pod Disruption Budget (PDB)?

Pointers

Defines minimum available pods during voluntary disruptions

Prevents too many pods from going down during drain/updates

✅ Scenario style
1. “When upgrading a deployment, I set PDB to minAvailable: 2 to ensure at least 2 pods are
always running so user traffic isn’t impacted.”

2. “We needed to upgrade the node pool for a production app. We set a PDB to minAvailable: 2
to keep at least 2 pods online even during the drain.”

PDB

+--> minAvailable: 2

+--> maxUnavailable: 1

Advanced Q&A
Q: Does PDB protect from node failures?
A: No, PDB only controls voluntary disruptions (e.g., drain, upgrades).

Best practices

always set PDB for critical deployments

test the effect by cordoning a node

🧠 Easy to remember: Minimum running pods during changes

Q. Shell script for factorial of a number

Q. Tell me about VPC structure in your project
Pointers

VPC CIDR
Public & private subnets
NAT gateway
Internet gateway
Route tables
Security groups

✅ Scenario style
“We designed a VPC with /16 CIDR, split into multiple private subnets for application servers,
with a NAT gateway for outbound internet, public subnets for the ALB, and separate security
groups for database layers.”

“Our VPC has a /16 CIDR block, split into public subnets for the ALB and private subnets for
EC2/EKS nodes. NAT gateways handle outbound traffic for private subnets. RDS is placed in
private subnet.”

Advanced Q&A
Q: Why put RDS in a private subnet?
A: For security — no direct internet access.

Best practices

separate public/private
follow least privilege on security groups

🧠 Easy to remember: CIDR, Subnets, NAT, IGW, Routing

Q. How is your CI/CD pipeline set up? What security tools are
integrated?
Pointers

Jenkins/GitLab
Docker builds
SonarQube (code scan)
Trivy/Anchore (image scan)
HashiCorp Vault (secrets)

Code → Build → Scan (Trivy) → Test → Deploy to K8s

✅ Scenario style
“Our pipeline is on GitLab CI, running Docker builds, security scanning with Trivy, static code
analysis with SonarQube, and uses Vault to inject secrets. This ensures secure, consistent,
automated releases.”

Advanced Q&A
Q: How do you manage secrets?
A: Vault or SSM Parameter Store, never hard-coded.

Best practices

integrate image scanning

automate secrets rotation

🧠 Easy to remember: Build, Scan, Store secrets, Deploy

Q. How do you manage them?

Pointers

Version control (Git)

IaC for infra
Role-based access
Automated tests

✅ Scenario style
“We manage pipelines through version-controlled YAML, infrastructure with Terraform, and
RBAC controls in Kubernetes to delegate least privilege.”

🧠 Easy to remember: Version, Automate, Secure

Q. Write a rough pipeline script for microservices

architecture
Advanced Q&A
Q: How do you handle dependencies?
A: Use semantic versioning + separate pipelines to avoid coupling.

Best practices

keep microservices decoupled tag images uniquely

🧠 Remember: Build → Test → Deploy

Q. What is multi-stage Docker build?

Pointers

Separate build & run stages

Reduce image size
Improves security

✅ Scenario style
“In a microservice build, we use a Golang builder image, compile binaries, and then copy them
to a scratch image in a second stage. That keeps the production image minimal.”

Advanced Q&A
Q: Why use multi-stage?
A: Reduce attack surface and image size.
Best practices

keep only production dependencies

scan final images

🧠 Easy to remember: Build once, copy artifacts, keep clean

Q. What are manifest files?

✅ Pointers
YAML files for Kubernetes resources
Define pods, deployments, services, etc.

✅ Scenario style
“We manage Kubernetes manifests for deployments and services in a GitOps workflow to
apply them consistently across environments.”

“We store Deployment and Service YAMLs in Git repos. We apply them with kubectl or
FluxCD.”

Advanced Q&A
Q: How to manage multiple environments?
A: Use Kustomize or Helm.

Best practices

version manifests
keep separate folders for dev/prod

🧠 Easy to remember: K8s blueprints

Q. What is Ansible Vault?

Pointers

Encrypt sensitive data (passwords, secrets)

Stored in playbooks securely
[ ansible-playbook ] --> [ encrypted vault file ]

✅ Scenario style
“We use Ansible Vault to encrypt DB passwords in our inventory, and decrypt only during
runtime with a vault password file.”

Advanced Q&A
Q: What if you lose the vault password?
A: You cannot decrypt — store vault password securely.

Best practices

rotate vault passwords

never commit vault keys to Git

🧠 Easy to remember: Encrypted secrets for playbooks

Q. How to make a K8s cluster highly available?

Pointers

Multiple control plane nodes

Etcd cluster quorum
Load balancer in front of control plane
Spread worker nodes across AZs

✅ Scenario style
“We deployed 3 control plane nodes with an external HA load balancer and spread worker
nodes in 3 AZs to achieve high availability.”

Advanced Q&A
Q: How do you handle etcd failure?
A: Ensure odd number of etcd members and frequent snapshots.

Best practices

HA LB in front of control planes

spread AZs
🧠 Easy to remember: Multi-master + Load Balancer + Spread

Q. Monitoring tools & common pod errors

Pointers

Prometheus, Grafana
Alertmanager
Common errors: CrashLoopBackOff, ImagePullBackOff

✅ Scenario style
“We use Prometheus + Grafana for metrics and Alertmanager for notifications. The most
common pod issue I handled was CrashLoopBackOff due to wrong configmaps or missing
secrets.”

Advanced Q&A
Q: What’s CrashLoopBackOff?
A: Container keeps crashing repeatedly, often due to bad configs.

Best practices

set up alerts
test alert receivers regularly

🧠 Easy to remember: Prometheus + Grafana + Alerts

Q. Terraform script for VPC architecture

Advanced Q&A
Q: How to handle state files?
A: Use remote backend with S3 + DynamoDB locking.

Best practices

use terraform fmt

version lock your providers

🧠 Remember: VPC → Subnets → IGW

Q. How many objects can an S3 bucket store?

✅ Answer
Unlimited objects
Practically limited by storage quotas

“S3 scales virtually unlimited. One bucket can store billions of objects.”

Advanced Q&A
Q: Any hard limits?
A: Only practical ones (like request rates), no hard object limit.

Best practices

enable bucket versioning

use lifecycle rules

🧠 Easy to remember: Unlimited

Q. What are IAM roles and policies?

Pointers

Roles: identities with permissions

Policies: permission documents in JSON
Used to grant granular access

✅ Scenario style
“I assigned an IAM role to an EC2 instance with a policy to only allow S3 access for backup
storage.”

Advanced Q&A
Q: Difference between policy and role?
A: Role = identity; Policy = permission rules.

Best practices

follow least privilege

audit IAM regularly

🧠 Easy to remember: Role = identity, Policy = rules

Q. What are artifacts?

Pointers

Build outputs
Stored in artifact repositories like Artifactory or S3
✅ Scenario style
“Our CI pipeline pushes JAR artifacts to Artifactory after a successful Maven build.”

“Our Maven builds produce JARs stored in Artifactory as versioned artifacts.”

Advanced Q&A
Q: Why store them?
A: To enable rollback or re-deploys.

Best practices

automate artifact cleanup

version artifacts clearly

🧠 Easy to remember: Build results

Q. SATS and DATS?

Common terms

SATS = System Acceptance Testing

DATS = Data Acceptance Testing

SATS → DATS

✅ Scenario style
“After deploying, we perform SATS to validate application behavior, and DATS to verify data
correctness with staging data.”

Advanced Q&A
Q: Are they manual or automated?
A: Usually manual with automated test cases integrated.

Best practices

always document acceptance criteria

automate where possible
🧠 Easy to remember: System + Data Testing

Q. How do you find errors in pipelines?

Pointers

Logs
CI/CD dashboard
Alerts
Test reports

✅ Scenario style
“Whenever the pipeline fails, I check Jenkins logs, review the failing stage, and correlate with
Git commit changes.”

Advanced Q&A
Q: How do you get notified?
A: Slack or email from pipeline notifications.

Best practices

break pipeline into small stages

add retry steps

🧠 Easy to remember: Logs + Stage + Commit

Q. What are Ansible Roles?

Pointers

Reusable set of tasks

Structured like modules
Makes playbooks cleaner

✅ Scenario style
“We created roles for installing Nginx, managing users, and deploying apps, to keep our
playbooks DRY and modular.”

Advanced Q&A
Q: Benefits of roles?
A: Reusable, maintainable, clean structure.

Best practices
structure roles with defaults, tasks, handlers
version control them

🧠 Easy to remember: Reusable Ansible blocks

Thanks Everyone!
Connect with me: Amit Singh