Harness Certification Interview Questions (2027)
Harness certification guide: Developer, Administrator and Architect exams, format, $50 paid exams, retakes, and 52 practice questions by exam domain.
Harness certifications come from Harness University at three levels: Developer (free, multiple choice, based on free plans), Administrator and Architect (paid, a knowledge exam plus a hands-on lab, based on paid plans). This page explains how the exams work and gives 52 practice questions mapped to the published CD & GitOps and CI exam domains.
All exam facts below come from the Harness University pages on the Harness Developer Hub, checked in November 2026. Harness changes its catalogue, so confirm details on the official study guide before you register. If you want deeper engineering questions rather than exam prep, read our advanced Harness interview questions, which is the main Harness interview guide on this site.
Key takeaways
- Developer certifications are free and use a 90-minute multiple choice exam. Harness lists Developer exams for 13 modules, including CD & GitOps, CI, Internal Developer Portal and Cloud & AI Cost Management.
- Administrator certifications exist for CD & GitOps, CI, Resilience Testing, Application Security Testing (STO) and Cloud & AI Cost Management. Architect certifications exist for CD & GitOps and CI.
- The CD & GitOps and CI Administrator and Architect pages list a price of $50, a 90-minute knowledge exam and a 120-minute hands-on exam.
- Exams are non-proctored and open book. Developer exams allow five retakes; paid exams allow two retakes per payment, then a 30-day cool-off.
- Badges are issued through Credly. The policies page does not state a validity period, passing score or question count.
- Each level builds on the one below it for the same module, so plan Developer, then Administrator, then Architect.
Harness certification levels at a glance
| Level | Based on | Exam format | Price (CD & CI pages) | Retakes |
|---|---|---|---|---|
| Developer | Free plans | Multiple choice knowledge exam, 90 minutes | Free | Five |
| Administrator | Paid plans | Knowledge exam (90 min) plus hands-on exam (120 min) | $50 | Two per payment, then 30-day cool-off |
| Architect | Paid plans | Knowledge exam (90 min) plus hands-on exam (120 min) | $50 | Two per payment, then 30-day cool-off |
Knowledge questions are delivered on Caveon Scorpion and the hands-on part runs in an Instruqt lab with a time limit. Prices for modules other than CD & GitOps and CI were not shown on the pages we checked, so look at each module's study guide.
Which modules have which levels
| Module | Developer | Administrator | Architect |
|---|---|---|---|
| Continuous Delivery & GitOps | Yes | Yes | Yes |
| Continuous Integration | Yes | Yes | Yes |
| Resilience Testing | Yes | Yes | No |
| Application Security Testing (STO) | Yes | Yes | No |
| Cloud & AI Cost Management | Yes | Yes | No |
| Code Repository, Internal Developer Portal, Infrastructure as Code Management, Database DevOps, Feature Management & Experimentation, API & Application Discovery, Application Security Testing (SCS), AI DLC Insights | Yes | No | No |
CD & GitOps exam domains
The Developer exam weights "Mastering Continuous Delivery" highest at 20%, followed by Kubernetes deployment and manifests (15%) and tagging and deployment strategies (15%). The Administrator exam spreads weight across eleven domains, with deployment management and troubleshooting and configuration management at 15% each. The Architect exam weights designing and implementing CD pipelines highest at 20%, with delegate management and IaC at 15% each. The questions below follow those domains.
Exam process questions (all levels)
1. What certification levels does Harness offer?
Developer, Administrator and Architect. Developer validates working knowledge of a module on the free plan, Administrator validates day-to-day administration on paid plans, and Architect validates design decisions and advanced configuration on paid plans. Administrator and Architect include a hands-on lab.
2. Is the Harness Developer certification free?
Yes. The Harness University policies page states that Developer certifications are free, and Administrator and Architect certifications are paid. The CD & GitOps and CI pages list $50 for each paid exam.
3. Are Harness exams proctored?
No. Harness describes the exams as non-proctored and open book, and you can take them at any time. Open book does not make them easy: the hands-on labs are timed, and you need to know where things are in the product to finish.
4. How many times can you retake an exam?
Free Developer exams allow five retakes. Paid Administrator and Architect exams allow two retakes per payment. After two retakes a 30-day cool-off applies, and each later retake also has a 30-day cool-off.
5. How is the hands-on part delivered?
On Instruqt, in a browser lab with a time limit. The policies page notes an idle prompt and an "Extend Time Limit" button. Read every task before you start so you can plan the order.
6. Where do badges come from?
Credly issues the digital badge, and you can download a PDF certificate from your Credly account. Add the badge link to your CV or LinkedIn profile so interviewers can check it.
7. Do Harness certifications expire?
The Harness University policies page does not state a validity period. Check your Credly badge for an expiry date and email [email protected] if you need a definite answer for an employer.
8. What are the prerequisites for the CD & GitOps Developer exam?
Harness lists basic terminal skills and a basic understanding of on-premises or cloud architecture. In practice you also need some Kubernetes knowledge, because 15% of the exam covers Kubernetes deployments and manifests.
CD & GitOps Developer practice questions
9. What is a service in Harness CD?
A service is the thing you deploy: its manifests or specification, its artifacts, and its variables. The same service can be deployed to many environments, which is why manifests use variables and overrides instead of hard-coded values.
10. What is the difference between an environment and an infrastructure definition?
An environment is a logical target such as QA or production, with an environment type of production or pre-production. An infrastructure definition is the concrete place inside it, such as a Kubernetes cluster and namespace. One environment can hold several infrastructure definitions.
11. Which manifest types can a Kubernetes service use?
Plain Kubernetes YAML, Helm charts, Kustomize and OpenShift templates, among others. The Developer exam expects you to know how a manifest is linked to an artifact so the image tag is filled in at deploy time.
12. What is a connector?
A connector stores how Harness reaches a third-party system, such as a Git provider, a Docker registry, a Kubernetes cluster or a cloud account, along with the credentials reference. Pipelines use connectors instead of embedding credentials.
13. Name the deployment strategies Harness supports for Kubernetes.
Rolling, canary and blue-green, plus basic for some deployment types. The exam objectives list basic, rolling, canary and blue-green. Know which one needs two copies of the application (blue-green) and which one gradually replaces pods (rolling).
14. How does a Harness Kubernetes canary work?
Harness deploys a canary workload alongside the current one, you verify it, a Canary Delete step removes it, and a rolling deployment updates the primary workload. Traffic split depends on pod counts unless you add a service mesh or ingress traffic routing.
15. How does a Harness Kubernetes blue-green swap traffic?
The Stage Deployment step deploys the new version behind the stage service, and the Swap Primary with Stage step switches the services so the new version receives production traffic. The swap does not health check for you, so add verification before it.
16. What is a trigger?
A trigger starts a pipeline automatically: a webhook from a Git event, a new artifact or manifest version, or a cron schedule. Triggers often reference an input set to supply runtime values.
17. What does an approval step do?
It pauses the pipeline until named users or groups approve or reject, or until it times out. Harness also supports Jira and ServiceNow approvals. Always set a timeout.
18. What is a failure strategy?
A rule that maps an error type to an action such as retry, rollback, abort or manual intervention. Failure strategies can be set at pipeline, stage, step group and step level, and the narrowest one applies.
19. What is Continuous Verification?
A step that compares metrics and logs from health sources for the new version against a baseline and fails the deployment if it detects anomalies. A verification failure can trigger rollback.
20. What is Harness GitOps built on?
Argo CD. You install a GitOps agent in the cluster, connect repositories and clusters, and define applications that sync from Git. The exam objectives include installing the GitOps agent.
21. What security controls does the Developer exam cover?
Role-based access control and two-factor authentication. Know that RBAC uses roles, resource groups and role bindings, and that permissions are additive.
CD & GitOps Administrator practice questions
22. How do you see why a deployment failed?
Open the execution, find the failed step, and read its logs and input and output tabs. For delegate issues, check delegate logs and capability checks. The Administrator exam puts 15% on deployment management and troubleshooting.
23. How do you ensure a step runs on a particular delegate?
Tag the delegate and add a delegate selector to the step, connector or pipeline. Only delegates with matching tags can take the task.
24. How do you install a delegate on Kubernetes?
Use the Helm chart or the Kubernetes manifest generated in the Harness UI, which includes the account ID and delegate token. The delegate connects outbound over HTTPS on port 443, so no inbound ports are needed.
25. How do Harness expressions reference a service variable?
With <+serviceVariables.name>. Expressions use the <+...> syntax and are evaluated with JEXL. The exam lists Harness expressions and variable management at 10%.
26. How do you override Helm values per environment?
Use overrides: environment-level or infrastructure-level values files and variables that take precedence over the service defaults. This keeps one service definition while QA and production use different replica counts or endpoints.
27. How do you add an artifact source from a private registry?
Create a connector to the registry with credentials stored as a secret, add the artifact source to the service, and set the tag as a runtime input or an expression from a trigger.
28. What is drift in GitOps and how do you detect it?
Drift is a difference between what Git declares and what runs in the cluster. Harness GitOps shows application sync status as out of sync when drift exists, and you can sync to restore the declared state.
29. How do you use an external secret manager?
Create a secret manager connector (for example HashiCorp Vault or AWS Secrets Manager), then create secrets that reference paths in it. Pipelines read them with <+secrets.getValue("id")>.
30. How do you block deployments during a release freeze?
Create a deployment freeze window scoped to the relevant organizations, projects, services or environments. Users with the override permission can still deploy, which is how emergency fixes pass.
31. How do you restrict a user group to pre-production?
Create a resource group containing only pre-production environments and bind a role with execute permission to that group. Production remains in a separate resource group.
32. What does a policy check on a pipeline do?
An OPA policy set attached to pipelines runs on events such as save or run and can warn or block when the pipeline breaks a rule, such as missing an approval before production.
33. How do you provision infrastructure in a CD stage?
Use Terraform or CloudFormation provisioner steps in the infrastructure section, so the target is created or updated before the deployment steps run. The Administrator exam weights IaC and provisioning at 10%.
CD & GitOps Architect practice questions
34. How do you make delegates highly available?
Run several replicas of the Kubernetes delegate across nodes or zones and size resources for peak load. Harness sends tasks to any healthy delegate matching the selector.
35. How do you apply least privilege across Harness?
Use narrow resource groups, separate service accounts for automation, scoped API tokens, delegate selectors that keep production credentials on production delegates, and secrets stored in an external manager.
36. Why use pipeline templates in a large organization?
Templates give every team the same approved stages and steps. Version labels let you roll out template changes gradually and pin production pipelines to a reviewed version.
37. How would you design rollback for a Helm deployment?
Use the native Helm rollback step in the rollback section, set failure strategies so verification failures roll back the stage, and keep database changes backward compatible so the old version can run against the new schema.
38. How do you deploy an umbrella Helm chart?
Define the parent chart as the service manifest, with subcharts as dependencies, and use values overrides per environment. The Architect objectives mention Helm and umbrella charts.
39. How do log and APM verification differ?
APM (metric) verification compares numeric series such as latency and error rate. Log verification clusters log messages and flags new or more frequent error patterns. Many teams use both.
40. How do you trigger a pipeline from a custom event?
Use a custom webhook trigger and call its URL with a payload from the external system. Steps read the payload through trigger expressions.
41. How do you scale CD to hundreds of services?
Template the pipeline, generate per-service pipelines with the Terraform provider or API, use artifact triggers, keep enough delegate capacity, and cap concurrency against shared infrastructure.
42. How do you deploy to AWS and Azure from one pipeline?
Create cloud connectors for each provider, define infrastructure definitions for each target (for example EKS and AKS), and use separate stages or a matrix over environments.
CI certification practice questions
The CI Developer exam spreads weight almost evenly across nine domains (12% for CI/CD concepts, 11% for each of the rest). The Administrator exam gives delegate configuration the largest single weight (16%). Harness lists the CI paid exams at $50 with the same 90-minute and 120-minute format.
43. What is a CI stage's build infrastructure?
Where the build steps run: Harness Cloud (hosted machines), a Kubernetes cluster, a local runner, or VMs. The choice affects cost, speed and network access.
44. What is a codebase in Harness CI?
The Git repository and branch or tag the pipeline clones, defined at pipeline level through a Git connector. Builds triggered by a webhook use the commit from the event.
45. How do Run steps work?
A Run step executes commands inside a container image you choose, such as maven or node. Pinning the image keeps builds reproducible.
46. How do you build and push a Docker image?
Use a Build and Push step for the target registry (Docker Hub, ECR, GCR or others) with a registry connector. It builds from a Dockerfile and pushes the tag you set.
47. How do you speed up CI builds?
Cache dependencies, run tests in parallel, and use Test Intelligence to run only tests affected by the change where it supports your language.
48. How do you share files between steps?
Steps in the same stage share a workspace. Between stages, save and restore caches or upload artifacts to storage.
49. Why does the CI Administrator exam cover delegates?
Kubernetes and VM build infrastructure need a delegate to start build pods or machines. Administrators must install, size and troubleshoot them, which is why the domain carries 16%.
Exam strategy and interview follow-up questions
50. How should you prepare for the hands-on lab?
Do every task in the study guide on a free Harness account, time yourself, and practise reading execution logs. The lab rewards knowing where settings live in the UI and YAML.
51. An interviewer asks what your Harness certification proves. How do you answer?
Name the level and module, then give one task you did in the lab and one place you used it at work. A badge plus a concrete example is stronger than the badge alone.
52. Which certification should a DevOps engineer take first?
CD & GitOps Developer, because it is free, covers the most common use of Harness, and is the base for the Administrator and Architect exams. Add CI Developer if your team builds in Harness too.
Practice lab tasks that match the exam objectives
Harness does not publish its lab tasks, so the list below is practice we built from the published Administrator and Architect objectives. Doing each one end to end on a free or trial account covers most of what the hands-on exams test.
| Practice task | Objective it maps to | What to check when done |
|---|---|---|
| Install a Kubernetes delegate with Helm, tag it, and route a connector to it with a selector | Delegate and infrastructure management | Connector test passes and the delegate shows as connected with the tag |
| Deploy a service with a canary strategy, add a Verify step, and force a failure | Deployment strategies, continuous verification | Verification fails and the failure strategy rolls the stage back |
| Create service variables and reference them in a manifest with expressions | Expressions and variable management | Rendered manifest in the execution log shows the resolved values |
| Add environment-level Helm values overrides for QA and production | Configuration management and parameter control | Replica counts differ per environment without changing the service |
| Connect a private registry and set the image tag as a runtime input | Artifact and repository management | Run form asks for the tag and the deployment uses it |
| Install a GitOps agent, create an application, then change the cluster by hand | GitOps and drift management | Application shows out of sync until you sync it |
| Create a Vault or cloud secret manager connector and read a secret in a step | Secret management | Value is masked in logs |
| Add a Terraform provisioning step that creates a namespace before deploying | IaC and provisioning | Infrastructure is created in the same run before the deploy step |
| Create a user group with access only to pre-production environments | User, access and identity management | A member cannot run the production stage |
| Create a deployment freeze and try to deploy during it | Policies, governance and deployment restrictions | The run is blocked unless the user has override permission |
| Turn a stage into a template with two versions and switch a pipeline between them | Designing CD pipelines (Architect) | The pipeline picks up the template change only when you change the version |
Common mistakes candidates make
- Studying only the knowledge exam. The Administrator and Architect labs are timed, and slow navigation costs more marks than missing facts.
- Skipping Kubernetes basics. Manifests, services, namespaces and Helm values show up across the CD domains.
- Treating open book as unlimited time. Looking up every answer in the docs will not fit in 90 minutes.
- Ignoring the domain weights. In the CD Developer exam, continuous delivery concepts carry 20%, while data visibility and security carry 5% each, so spend study time to match.
- Learning FirstGen terms. Current Harness uses pipelines, stages and steps; "workflows" is older FirstGen language and will confuse your answers.
- Not reading the study guide again before booking. Harness updates objectives and product names, such as Cloud Cost Management becoming Cloud & AI Cost Management.
Study plan
- Week 1: read the CD & GitOps Developer study guide, create a free account, deploy a sample app to a local Kubernetes cluster.
- Week 2: practise rolling, canary and blue-green, triggers, approvals and a GitOps application, then sit the Developer exam.
- Weeks 3 to 4: work through the Administrator objectives: delegates, expressions, overrides, secret managers, freezes, RBAC and OPA.
- Weeks 5 to 6: for Architect, design a templated multi-service pipeline with verification and Terraform provisioning.
Kubernetes skills carry much of the CD exam. Our CKA training in Pune and DevOps training (Kharadi classroom and live online) build that base. For interview rounds, pair this page with our Harness CD interview questions, the Harness engineer guide and scenario-based Harness questions.
Official references
- Harness University
- CD & GitOps certifications and study guides
- CI certifications and study guides
- Harness University policies and FAQs
FAQs
How much does a Harness certification cost?
Developer certifications are free. The CD & GitOps and CI Administrator and Architect pages list $50 per exam. Check the study guide for other modules before registering.
Are Harness certification exams proctored?
No. Harness describes them as non-proctored and open book, available at any time. Paid exams include a timed hands-on lab on Instruqt.
How long is the Harness CD Developer exam?
90 minutes of multiple choice questions. The Administrator and Architect exams add a 120-minute hands-on exam to a 90-minute knowledge exam.
How many retakes do I get?
Five for free Developer exams. Two per payment for Administrator and Architect exams, followed by a 30-day cool-off before each further attempt.
Which Harness modules have an Architect certification?
CD & GitOps and Continuous Integration. Resilience Testing, STO and Cloud & AI Cost Management go up to Administrator.
Do Harness certifications expire?
The Harness University policies page does not give a validity period. Check your Credly badge or ask [email protected].
Is a Harness certification useful for DevOps jobs in India?
It helps when a job description names Harness, because it shows you have used the product. Interviewers still test Kubernetes, Git and pipeline fundamentals, so pair it with hands-on practice.
What's Your Reaction?
Like
0
Dislike
0
Love
0
Funny
0
Angry
0
Sad
0
Wow
0