Argo CD Interview Prep Guide: 4-Week Plan (2027)
Four-week Argo CD interview prep plan: install, core concepts, declarative Applications, auto-sync, sync waves, rollback, app of apps and ApplicationSets.
To prepare for an Argo CD interview, install Argo CD in a test cluster, deploy an app from Git with the CLI and then declaratively, practise auto-sync, pruning, self-heal, sync waves and rollback, and learn the core vocabulary: Application, target state, live state, sync status and health. This guide turns that into a four-week plan with checkpoints.
This is a preparation guide for people starting out with Argo CD or returning to it. It does not repeat the question banks. When you finish the plan, move on to the scenario-based Argo CD interview questions, which is the main post in this group, and the 70+ Argo CD interview questions for breadth.
Key takeaways
- Argo CD v3.6.1 was the newest release in November 2026. The docs recommend a pinned version for production instead of the unpinned stable manifests.
- Learn the vocabulary first. Interviewers expect exact meanings for target state, live state, sync status, health and refresh.
- Automated sync does not prune or self-heal unless you enable
prune: trueandselfHeal: true. - Without the
resources-finalizer.argocd.argoproj.iofinalizer, deleting an Application does not delete the resources it manages. - Sync waves order resources with the
argocd.argoproj.io/sync-waveannotation; resources without one are in wave 0.
The four-week plan at a glance
| Week | Topic | Checkpoint |
|---|---|---|
| 1 | GitOps ideas, Argo CD architecture and vocabulary, install | Argo CD running in kind or minikube, guestbook deployed from the CLI |
| 2 | Declarative Applications, AppProjects, auto-sync options | An Application YAML in Git that syncs itself, with prune and self-heal on |
| 3 | Sync waves, hooks, Helm and Kustomize sources, rollback | A database migration Job that runs before the app, and a working rollback |
| 4 | App of apps, ApplicationSets, RBAC, SSO, rehearsal | One root Application that creates three child apps |
Week 1: concepts and first deploy
Start with the GitOps idea: Git holds the desired state, and an agent in the cluster keeps the cluster matching it. If that is new to you, read what GitOps is and how it improves infrastructure automation before you install anything.
Then learn these terms from the Argo CD core concepts page. You will use them in almost every answer.
| Term | Meaning in the Argo CD docs |
|---|---|
| Application | A group of Kubernetes resources as defined by a manifest; it is a custom resource |
| Target state | The desired state of an application, as represented by files in Git |
| Live state | What is actually running in the cluster |
| Sync status | Whether the live state matches the target state |
| Sync | The process of moving an application to its target state |
| Health | Whether the application is running correctly and can serve requests |
| Refresh | Comparing the latest code in Git with the live state |
Now install it. The getting started guide uses these steps:
kubectl create namespace argocd
kubectl apply -n argocd --server-side --force-conflicts \
-f https://raw.githubusercontent.com/argoproj/argo-cd/stable/manifests/install.yaml
argocd admin initial-password -n argocd
argocd login <ARGOCD_SERVER>
argocd app create guestbook \
--repo https://github.com/argoproj/argocd-example-apps.git \
--path guestbook \
--dest-server https://kubernetes.default.svc \
--dest-namespace default
argocd app sync guestbook
The initial admin password lives in the argocd-initial-admin-secret secret, and the docs say to delete that secret after you change the password. Note that the install above tracks the stable branch; for anything beyond a lab, pin a release tag as the docs recommend.
Checkpoint questions. What is the difference between sync status and health? (Sync status compares Git with the cluster; health asks whether the app works. An app can be Synced and Degraded.) What does a refresh do that a sync does not? (It compares Git with the cluster without applying anything.)
Week 2: declarative Applications and auto-sync
Clicking in the UI is fine for learning, but interviewers want to hear that Applications themselves live in Git. Write the guestbook app as YAML:
apiVersion: argoproj.io/v1alpha1
kind: Application
metadata:
name: guestbook
namespace: argocd
finalizers:
- resources-finalizer.argocd.argoproj.io
spec:
project: default
source:
repoURL: https://github.com/argoproj/argocd-example-apps.git
targetRevision: HEAD
path: guestbook
destination:
server: https://kubernetes.default.svc
namespace: guestbook
syncPolicy:
automated:
prune: true
selfHeal: true
Three details from the docs are common interview points. The Application must be created in the Argo CD namespace, argocd by default. The finalizer makes deletion cascade to the managed resources. And automated sync is conservative by default: it does not delete resources removed from Git unless prune is true, and it does not revert manual changes in the cluster unless selfHeal is true.
Test both. Delete a Deployment from Git and watch it disappear from the cluster. Then run kubectl scale on a Deployment by hand and watch Argo CD put the replica count back. The docs say self-heal retries after a timeout that defaults to 5 seconds, set by --self-heal-timeout-seconds on the application controller.
Also learn AppProjects. The docs describe a project as a logical grouping of applications that limits which repositories they can pull from, which clusters and namespaces they can deploy to, and which roles apply. Create a project for one team and try to deploy outside its namespace to see the error.
Checkpoint questions. Why is allowEmpty off by default? (So a broken path or empty commit cannot prune every resource.) How do you pause auto-sync without deleting the policy? (Set spec.syncPolicy.automated.enabled to false; the docs say the controller then skips automated sync even if prune and self-heal are set.)
Week 3: ordering, hooks, templates and rollback
Real deployments need order. Add the argocd.argoproj.io/hook: PreSync annotation to a database migration Job, and sync-wave annotations to the resources that depend on each other.
metadata:
annotations:
argocd.argoproj.io/hook: PreSync
---
metadata:
annotations:
argocd.argoproj.io/sync-wave: "-1"
The docs give the ordering rules: by phase (PreSync, Sync, PostSync, with SyncFail running when a sync fails), then by wave from lowest to highest, then by kind (namespaces first, then other Kubernetes resources, then custom resources), then by name. Resources with no wave are in wave 0, and negative waves run first.
Next, point an Application at a Helm chart and at a Kustomize overlay so you can explain both sources. Then practise recovery with the CLI:
argocd app diff guestbook
argocd app history guestbook
argocd app rollback guestbook <HISTORY_ID>
argocd app wait guestbook --health
Know the catch with rollback: with automated sync on, Argo CD will move the app forward again to whatever Git says. In a GitOps team, the usual fix is to revert the commit in Git. Use CLI rollback for emergencies, and be ready to explain how you would turn auto-sync off first.
Checkpoint questions. Where does a PostSync hook run? (After all Sync hooks completed successfully, the application is applied and all resources are Healthy.) What does argocd app diff compare? (Target state with live state.)
Week 4: scaling up and security
- App of apps. One root Application whose path contains other Application manifests. The docs describe this as a way to deploy and configure a group of apps together.
- ApplicationSets. A controller that generates Applications from generators, such as a fixed list of values or the clusters registered in Argo CD. Use it when many apps or clusters share one template. Our advanced Argo CD questions cover ApplicationSets and high availability in depth.
- RBAC and SSO. Map SSO groups to project roles so teams can sync their own apps and nothing else. See the Argo CD security FAQs for the details.
- Secrets. Never commit plain Kubernetes Secrets. Be ready to describe one approach your team would use, such as an external secrets operator or encrypted secrets in Git.
Finish the week by explaining your lab out loud in five minutes: what is in Git, what Argo CD does on each commit, how you would roll back, and who is allowed to sync.
Build a lab repository you can show
Keep everything from the four weeks in one Git repository: the Application and AppProject manifests, a Helm chart or Kustomize overlay, the PreSync migration Job and a short README that explains how to bootstrap it. In an interview, walking through a real repository is more convincing than describing one, and it gives you concrete answers for questions about structure, review and promotion between environments.
Questions to answer before your interview
- What is Argo CD? A declarative GitOps continuous delivery tool for Kubernetes that keeps the cluster matching the state in Git.
- Push vs pull deployment? A CI pipeline pushes with cluster credentials; Argo CD runs inside the cluster and pulls from Git, so CI does not need cluster access.
- What does OutOfSync mean? The live state differs from the target state in Git.
- Synced but Degraded: what now? The manifests applied but the app is unhealthy. Check pod events and logs; the problem is in the app or its config, not in Argo CD.
- What do prune and selfHeal do? Prune deletes resources removed from Git; selfHeal reverts manual changes in the cluster. Both are off unless enabled.
- What does the resources finalizer do? It makes deleting an Application also delete its managed resources.
- How do you run a migration before the app updates? A Job with the PreSync hook annotation.
- How do you order resources within a sync? Sync-wave annotations; lower waves apply first and the default is 0.
- What is an AppProject? A grouping of Applications that restricts source repositories, destinations and roles.
- App of apps or ApplicationSet? App of apps for a handful of hand-written apps; ApplicationSet when apps are generated from a pattern across many clusters.
- How do you roll back? Revert in Git when auto-sync is on.
argocd app rollbackworks for emergencies once auto-sync is paused. - How do you upgrade Argo CD safely? Read the upgrade notes, pin the new version, apply it in a non-production cluster first, then production.
How this fits a DevOps career
Argo CD questions usually appear in platform, SRE and Kubernetes roles, alongside CI and cluster administration. The DevOps training covers GitOps with CI/CD, containers and cloud, and the CKA training builds the cluster skills Argo CD depends on. Both run in our Kharadi, Pune classroom and live online across India.
Common mistakes in Argo CD interviews
- Saying auto-sync deletes removed resources by default. It does not unless
pruneis on. - Using CLI rollback with auto-sync still on, then wondering why the app moves forward again.
- Describing the Application finalizer backwards. Without it, deleting the Application leaves the resources running.
- Mixing up refresh and sync. A refresh compares; a sync applies.
- Installing from the unpinned stable manifests in production. Pin a release tag and upgrade on purpose.
Each of these comes up because the default behaviour surprised someone. If you can explain why Argo CD behaves this way, you show that you have run it rather than read about it.
Frequently asked questions
How long does it take to prepare for an Argo CD interview?
About four weeks at an hour a day if you already know basic Kubernetes. Spend the first two weeks on install, vocabulary and declarative Applications.
What is the difference between sync status and health in Argo CD?
Sync status shows whether the live state matches Git. Health shows whether the application is running correctly. An app can be Synced and still Degraded.
Does Argo CD auto-sync delete resources by default?
No. Automated sync only deletes resources removed from Git when prune: true is set, and only reverts manual cluster changes when selfHeal: true is set.
How do I get the initial Argo CD admin password?
Run argocd admin initial-password -n argocd. The password is stored in the argocd-initial-admin-secret secret, which you should delete after changing it.
What is the default sync wave?
Wave 0. Resources and hooks without the argocd.argoproj.io/sync-wave annotation are in wave 0, and negative waves run earlier.
Which Argo CD version should I learn?
The current release line. v3.6.1 was the newest release in November 2026, and the docs recommend pinning a version for production.
References
What's Your Reaction?
Like
0
Dislike
0
Love
0
Funny
0
Angry
0
Sad
0
Wow
0