Real-Time Argo CD Interview Questions: CLI and Ops (2027)

52 hands-on Argo CD interview questions: argocd app commands, diff, sync, rollback, RBAC checks, notifications, metrics and live troubleshooting drills.

Sep 17, 2025 - 11:56
Nov 10, 2026 - 09:38
 0  114
Real-Time Argo CD Interview Questions: CLI and Ops (2027)

Real-time Argo CD interview questions test whether you can operate Argo CD on a live cluster: read sync and health status, run the right argocd app commands, find why a sync failed, roll back safely, check RBAC and watch the controllers through logs and metrics. This page has 52 hands-on questions grouped by experience level, plus live troubleshooting drills.

Commands and metric names were checked against the Argo CD documentation in November 2026. If you need the concepts first, start with our 70+ Argo CD interview questions. For design-style problems such as multi-cluster layout and secrets strategy, see the scenario-based Argo CD questions. This page sits between the two: what you type and what you read when Argo CD is running in front of you.

Key takeaways

  • Interviewers for "real-time" rounds often share a terminal or the Argo CD UI and ask you to explain what you see, so practise reading argocd app get output aloud.
  • Know the two status fields cold: sync status (Synced, OutOfSync, Unknown) and health status (Healthy, Progressing, Degraded, Suspended, Missing, Unknown).
  • The commands that come up most are get, diff, sync, wait, history, rollback, logs, manifests and terminate-op.
  • The application controller exposes metrics on port 8082, the API server on 8083 and the repo server on 8084. argocd_app_info, argocd_app_sync_total and argocd_app_reconcile are the metrics to name.
  • Rollback through Argo CD is blocked while automated sync is enabled. Saying this unprompted shows real use.
  • For live drills, talk through symptom, where you look, the fix and how you stop it happening again.

How this page is organised

SectionQuestionsWhat it checks
Freshers1 to 18Logging in, reading status, basic CLI
2 to 5 years19 to 36Diffs, partial syncs, hooks, RBAC checks, notifications
Senior37 to 44Metrics, controller performance, upgrades, operating at scale
Live troubleshooting drills45 to 52Step-by-step answers to problems shown on screen

Argo CD is a GitOps tool. If GitOps itself is new to you, read what GitOps is and how it helps infrastructure automation before the questions.

Real-time Argo CD questions for freshers

1. How do you log in to Argo CD from the CLI?

Run argocd login <server-address> and enter a username and password, or add --sso if the server uses single sign-on. On a fresh install the user is admin, and the initial password is stored in the argocd-initial-admin-secret Secret in the argocd namespace. You can print it with argocd admin initial-password -n argocd. Change it, then delete that Secret.

argocd admin initial-password -n argocd
argocd login argocd.example.com
argocd account update-password

2. How do you check which Argo CD server and user the CLI is using?

Use argocd context to list saved server contexts and switch between them, and argocd account get-user-info to see the logged-in user and groups. argocd version shows both client and server versions, which matters when a command behaves differently from the docs.

3. How do you list applications and see their state at a glance?

argocd app list prints every application you can see, with its project, sync status, health status, sync policy and source. Add -p <project> to filter by project or -l key=value to filter by label.

4. What does argocd app get show?

It shows one application's source repo, path and revision, destination cluster and namespace, sync policy, sync status, health status, the last operation and a table of every managed resource with its own status. Add -o yaml to see the full Application object, including status.conditions, where errors such as comparison failures appear.

5. What is the difference between sync status and health status?

Sync status says whether the live objects match what Git describes. Health status says whether those objects are working. An app can be Synced and Degraded (the manifests applied, but the pods crash) or OutOfSync and Healthy (someone changed a replica count by hand and the app still runs).

6. Name the health states Argo CD reports.

Healthy, Progressing, Degraded, Suspended, Missing and Unknown. Progressing is normal during a rollout. Missing means the resource is in Git but not in the cluster. Suspended appears for things like a paused rollout or a suspended CronJob.

7. How do you trigger a sync from the CLI?

argocd app sync <app>. Useful flags: --prune to delete resources removed from Git, --dry-run to preview, --revision to sync a specific commit, and --async to return without waiting.

8. How do you wait until an app is healthy, for example in a pipeline?

argocd app wait <app> --sync --health --timeout 300 blocks until the app is synced and healthy or the timeout expires. It returns a non-zero exit code on timeout, so a CI job fails clearly instead of hanging.

9. What is the difference between refresh and hard refresh?

A refresh compares Git with the live state again. A hard refresh also clears the cached manifests so the repo server regenerates them, which you need after changing a Helm chart dependency or a plugin output that the cache would otherwise hide. From the CLI: argocd app get <app> --refresh or --hard-refresh.

10. How often does Argo CD check Git if no webhook is configured?

By default it polls every 180 seconds (3 minutes), set by timeout.reconciliation in the argocd-cm ConfigMap, with some jitter added. A Git webhook pointed at the API server's /api/webhook endpoint makes changes show up almost immediately.

11. How do you see the history of deployments for an app?

argocd app history <app> lists past syncs with an ID, date and Git revision. You need the ID for a rollback.

12. How do you view logs of the pods an app runs?

argocd app logs <app> streams logs from the app's pods. You can narrow it with --kind, --name, --container and -f to follow. It saves switching to kubectl and guessing pod names.

13. How do you list every resource an application manages?

argocd app resources <app>. It shows group, kind, namespace, name and whether each resource is orphaned or needs pruning.

14. How do you print the manifests Argo CD would apply?

argocd app manifests <app> prints the rendered target manifests from Git. With --source live it prints what is actually in the cluster. Comparing the two is a quick way to understand a diff.

15. How do you see which clusters and repositories are registered?

argocd cluster list shows clusters with their connection status and server version. argocd repo list shows repositories and whether Argo CD can connect to them. A failed repo connection explains many "ComparisonError" messages.

16. How do you stop a sync that is stuck?

argocd app terminate-op <app> stops the running operation. Then find the cause, often a hook Job that never finishes or a resource waiting on a health check, before you sync again.

17. How do you create a simple application from the CLI?

argocd app create guestbook \
  --repo https://github.com/argoproj/argocd-example-apps.git \
  --path guestbook \
  --dest-server https://kubernetes.default.svc \
  --dest-namespace default

In real teams the Application is usually declared as YAML in Git, but the CLI form is common in interviews because it shows you know the required fields: repo, path, destination server and namespace.

18. Where would you look first if the UI shows a red "ComparisonError"?

Run argocd app get <app> -o yaml and read status.conditions. Common causes are a repo Argo CD cannot reach, a wrong path or branch, or a Helm or Kustomize render error. The repo server logs show the full render error.

Real-time Argo CD questions for 2 to 5 years of experience

19. How do you read argocd app diff output?

It prints a unified diff per resource between the live object and the target from Git. Lines starting with - are live values that would change and + lines are what Git wants. The command exits with code 1 when there is a difference, which is handy in scripts and pull request checks.

20. How do you sync only one resource?

Use --resource GROUP:KIND:NAME, for example argocd app sync myapp --resource apps:Deployment:web. For core resources the group is empty, as in :ConfigMap:web-config. Partial syncs are useful during an incident, but remember the rest of the app stays OutOfSync.

21. A field keeps showing as different even though nobody changed it. What do you do?

First confirm which controller or webhook writes the field (a mutating webhook, an HPA, a defaulting operator). Then add an ignoreDifferences entry for that field in the Application spec. If automated sync is on, also set the RespectIgnoreDifferences=true sync option so the sync does not overwrite it.

spec:
  ignoreDifferences:
  - group: apps
    kind: Deployment
    jsonPointers:
    - /spec/replicas
  syncPolicy:
    syncOptions:
    - RespectIgnoreDifferences=true

22. How do you switch an app from automated to manual sync during an incident?

argocd app set <app> --sync-policy none. This lets you hold a hotfix or roll back without Argo CD putting the Git state back. Record it, and turn automation back on once Git reflects the fix. If the Application itself is managed from Git (app of apps), change it there or the parent will revert your edit.

23. How do you roll back, and what blocks a rollback?

argocd app rollback <app> <history-id>; without an ID it goes to the previous version. Argo CD refuses a rollback while automated sync is enabled, because the next reconcile would sync back to Git. The cleaner GitOps answer is to revert the commit in Git and let Argo CD sync it.

24. How do you check what a user is allowed to do?

A user can run argocd account can-i sync applications 'myproject/*'. An admin can test a policy file before applying it with argocd admin settings rbac can <role> <action> <resource> <object> --policy-file policy.csv. Testing this way avoids locking a team out after a change to argocd-rbac-cm.

25. How do you see why a hook failed?

Run argocd app get <app> and find the hook resource in the list, usually a Job, with its phase and message. Then read the Job's pod logs with argocd app logs <app> --kind Job --name <job>. If the hook has a delete policy of HookSucceeded, a failed hook stays around for inspection, while a successful one is removed.

26. How do you check sync windows from the CLI?

argocd proj windows list <project> shows allow and deny windows with their schedule, duration and which apps or clusters they cover. When a sync is blocked, the app shows a message saying a sync window is active.

27. How do you get alerts when a sync fails?

Use Argo CD Notifications. Configure a service such as Slack in argocd-notifications-cm, keep the token in argocd-notifications-secret, and subscribe an app with an annotation:

metadata:
  annotations:
    notifications.argoproj.io/subscribe.on-sync-failed.slack: deploy-alerts

The bundled catalog includes triggers such as on-sync-failed, on-health-degraded and on-deployed.

28. How do you force a hard refresh without the CLI?

Click Refresh, then Hard Refresh in the UI, or annotate the Application with argocd.argoproj.io/refresh: hard. Argo CD removes the annotation after it acts on it.

29. Where do you look for controller errors?

The application controller runs as a StatefulSet, so use kubectl -n argocd logs statefulset/argocd-application-controller. Render problems show up in kubectl -n argocd logs deploy/argocd-repo-server, and login or API errors in the argocd-server logs.

30. How do you test a change to a Helm values file before merging?

Point a temporary app or argocd app diff at the branch with --revision <branch>, or render locally with helm template using the same values. Some teams post the diff on the pull request. The point is to see the rendered change, not just the values change.

31. What do sync operation phases tell you?

The last operation has a phase: Running, Succeeded, Failed, Error or Terminating. Failed means a resource could not be applied or a hook failed. Error usually means Argo CD itself could not complete the operation, for example because manifests failed to render.

32. How do you find orphaned resources?

Enable orphaned resource monitoring on the AppProject (spec.orphanedResources). Argo CD then flags resources in the app's namespace that no app manages, and the UI can show them. Set warn: true to get a condition on the app as well.

33. A deployment uses an image tag that changed in the registry, but Argo CD shows Synced. Why?

Argo CD compares manifests, not registry contents. If the tag string in Git did not change, there is nothing to sync. Use immutable tags or digests, or Argo CD Image Updater to write new tags back to Git.

34. How do you run a sync that replaces rather than applies a resource?

Use the Replace=true sync option on that resource or the app. It is needed for objects too large for client-side apply annotations, but it deletes and recreates where it must, so use it carefully. ServerSideApply=true is another option for large objects.

35. How do you check an app's events?

The UI resource view has an Events tab. From the terminal, kubectl -n <namespace> get events --sort-by=.lastTimestamp works on the target cluster. Argo CD also records its own Kubernetes events on the Application object in the argocd namespace.

36. How do you add a cluster and confirm it works?

argocd cluster add <kube-context> creates a service account on the target and stores the credentials as a Secret. Then argocd cluster list should show the connection as Successful once an app targets it. Declarative teams create the cluster Secret with the label argocd.argoproj.io/secret-type: cluster instead.

Real-time Argo CD questions for senior engineers

37. Which Argo CD metrics would you put on a dashboard?

Scrape the application controller (argocd-metrics:8082/metrics), API server (argocd-server-metrics:8083/metrics) and repo server (argocd-repo-server:8084/metrics). Plot argocd_app_info grouped by sync and health status, argocd_app_sync_total by phase, and the argocd_app_reconcile histogram for reconcile duration. Add repo server request latency and Git request errors.

38. Write an alert for apps that stay degraded.

- alert: ArgoAppDegraded
  expr: argocd_app_info{health_status="Degraded"} == 1
  for: 15m
  labels:
    severity: warning
  annotations:
    summary: "Argo CD app {{ $labels.name }} degraded for 15 minutes"

The for clause avoids paging on normal rollouts that pass through a bad state briefly. Our Prometheus interview questions cover alert rules in more depth.

39. Reconciliation is slow. What do you check?

Look at the argocd_app_reconcile histogram and controller CPU and memory. Then check the controller's status and operation processor counts, repo server replicas and cache hit rates, and whether many apps share one monorepo so every commit refreshes all of them. Fixes include more repo server replicas, controller sharding, and the manifest-generate-paths annotation to limit refreshes.

40. How does controller sharding work?

You run several application controller replicas and Argo CD splits clusters across them, so each shard reconciles a subset of clusters. Set the replica count and the ARGOCD_CONTROLLER_REPLICAS environment variable to match. Sharding is by cluster, so one very large cluster still lands on one shard.

41. How do you upgrade Argo CD with low risk?

Read the upgrade notes for every minor version you cross, back up with argocd admin export, upgrade a non-production instance first, and watch sync and health metrics after the change. Check RBAC and resource tracking changes in particular, because major versions have changed defaults.

42. How do you audit who synced what?

Argo CD records Kubernetes events on Application objects with the user and action, and the API server logs requests. Ship both to your log platform. Git history covers the desired state side, so you can answer "who changed it" and "who deployed it" separately.

43. How would you limit the blast radius of an admin mistake?

Use AppProjects to restrict source repos, destinations and allowed resource kinds; keep cluster-scoped resources in a separate platform project; give teams project roles instead of global roles; and protect production with sync windows and manual sync. Disable the local admin user once SSO works.

44. How do you explain Argo CD's own resource usage to a capacity review?

Report app and resource counts, reconcile duration, controller memory (which grows with the number of watched resources), repo server CPU during renders, and Redis memory for the cache. Tie each to a scaling action: shard controllers, add repo servers, or trim watched resource kinds with exclusions.

Live troubleshooting drills

These mimic a shared-screen round. Each answer gives the order you would check things in.

45. The app is stuck in Progressing for 20 minutes after a sync.

Run argocd app get and find which resource is Progressing. For a Deployment, check rollout status and pod events: image pull errors, failing readiness probes or pending pods from lack of capacity. Fix the cause in Git. If a custom resource never reports health, add a custom health check in Lua in argocd-cm.

46. Sync succeeded, health is Degraded.

The manifests applied, so look at the workload: argocd app logs for crash output, pod events for OOMKilled or probe failures, and recent config changes in the diff of the last revision. Roll back by reverting in Git if users are affected, then debug.

47. The app flips between Synced and OutOfSync every few minutes.

Something in the cluster keeps changing a field. Use argocd app diff during the OutOfSync window to see which field. Typical culprits are an HPA on replicas, a mutating webhook adding sidecars or labels, or two apps managing the same resource. Fix with ignoreDifferences or by removing the duplicate owner.

48. A new app shows ComparisonError: repository not accessible.

Check argocd repo list. Then confirm the credentials Secret, the URL format (HTTPS vs SSH must match the credentials), known hosts for SSH, and network egress from the repo server pod. Test with argocd repo add using the same values to see the exact error.

49. A developer says "Argo CD deleted my resource".

Check the app's history and events to see if a sync with prune ran, and which commit removed the manifest. If it was intended, explain. If not, restore it in Git. To prevent repeats, add the Prune=false annotation to critical resources or use PrunePropagationPolicy and require manual sync in production.

50. The UI is very slow to load the application list.

Check API server CPU and logs, Redis health, and the number of apps. Very large app lists benefit from more API server replicas and from users filtering by project. Also check whether an ingress or proxy is timing out gRPC-web calls.

51. A sync hangs on a PreSync hook.

Find the hook Job in argocd app get, read its pod logs, and check whether it waits on something that does not exist yet, such as a database Service created in a later wave. Terminate the operation, fix the order with sync waves, and sync again.

52. After an upgrade, a team gets "permission denied" on actions that used to work.

Compare the RBAC policy with the upgrade notes for the version, since defaults for some resources and actions have changed across major versions. Test the role with argocd admin settings rbac can, update argocd-rbac-cm in Git, and confirm with the team.

Argo CD command cheat sheet

TaskCommand
Log inargocd login <server> [--sso]
List appsargocd app list -p <project>
Inspect one appargocd app get <app> [-o yaml]
Compare Git and clusterargocd app diff <app>
Preview a syncargocd app sync <app> --dry-run
Sync one resourceargocd app sync <app> --resource apps:Deployment:web
Wait in CIargocd app wait <app> --sync --health --timeout 300
Deployment historyargocd app history <app>
Roll backargocd app rollback <app> <id>
Stop a stuck operationargocd app terminate-op <app>
Pod logsargocd app logs <app> -f
Live vs target manifestsargocd app manifests <app> --source live
Disable auto-syncargocd app set <app> --sync-policy none
Check own permissionsargocd account can-i sync applications '<project>/*'
Test an RBAC policyargocd admin settings rbac can <role> <action> <resource> <object> --policy-file policy.csv
Clusters and reposargocd cluster list, argocd repo list
Sync windowsargocd proj windows list <project>

Mistakes that cost points in live rounds

  • Running kubectl edit to fix production while auto-sync is on, then being surprised when Argo CD reverts it. Say how you would pause auto-sync or fix it in Git.
  • Syncing with --prune before reading the diff. Always run argocd app diff or --dry-run first on a shared environment.
  • Treating Synced as "working". Check health and the workload itself.
  • Guessing at RBAC. Use can-i or admin settings rbac can to prove the answer.
  • Forgetting the Git side. Every fix you make by hand should end with a commit, or the next reconcile undoes it.
  • Restarting the controller as a first step. Read its logs and metrics before you restart anything.

How to practise for a live Argo CD round

  1. Install Argo CD on a local cluster (kind or minikube) and deploy the public guestbook example.
  2. Break it on purpose: wrong image tag, failing probe, manual kubectl edit, a deleted manifest. Fix each from the CLI and say aloud what you check.
  3. Add Prometheus and scrape the three metrics endpoints, then build one panel per metric named above.
  4. Set up one notification trigger to a test Slack channel or webhook.

Kubernetes skills carry most of these answers. Our Kubernetes interview questions and the GitOps tools comparison help fill gaps. If you want guided labs, the CKA training (classroom in Kharadi, Pune, or live online) covers the cluster side, and the 200+ DevOps interview questions hub links every other topic.

FAQ

What does "real-time" mean in an Argo CD interview?

It usually means hands-on questions about running Argo CD: commands, reading status, fixing failed syncs and rollbacks, often on a shared screen. Design questions are more common in scenario rounds.

Which Argo CD CLI commands should I memorise?

login, app list, app get, app diff, app sync, app wait, app history, app rollback, app logs, app manifests, app terminate-op, cluster list, repo list and account can-i.

Why does Argo CD refuse to roll back my app?

Rollback is not allowed while automated sync is enabled, because Argo CD would sync back to Git. Disable auto-sync or, better, revert the commit in Git.

What is the difference between refresh and hard refresh?

Refresh compares Git and the cluster again. Hard refresh also clears cached rendered manifests so they are generated again.

Which ports expose Argo CD metrics?

The application controller uses 8082, the API server metrics service 8083, and the repo server 8084, each at the /metrics path.

How do I get notified of failed syncs?

Use Argo CD Notifications: configure a service in argocd-notifications-cm and subscribe apps with an annotation such as notifications.argoproj.io/subscribe.on-sync-failed.slack.

How is this page different from other Argo CD question lists?

It focuses on operating commands and live troubleshooting. The 70+ questions page covers concepts and the scenario page covers design problems.

References

What's Your Reaction?

Like Like 0
Dislike Dislike 0
Love Love 0
Funny Funny 0
Angry Angry 0
Sad Sad 0
Wow Wow 0
Mridul I am a passionate technology enthusiast with a strong focus on DevOps, Cloud Computing, and Cybersecurity. Through my blogs at DevOps Training Institute, I aim to simplify complex concepts and share practical insights for learners and professionals. My goal is to empower readers with knowledge, hands-on tips, and industry best practices to stay ahead in the ever-evolving world of DevOps.