Interview question bank¶
Answer aloud in 60–120 seconds. Use a concrete example and state trade-offs.
Helm fundamentals¶
- What happens from
helm upgradeuntil Pods become Ready? - Explain chart version versus
appVersionversus image tag. - Explain values precedence, including multiple
-ffiles and--set. - How do
withandrangechange dot, and when do you use$? - Why use
includeinstead oftemplatein many helper patterns? - What do
required,tpl, andlookupdo, and what risks do they introduce? - What is stored for a Helm release and where?
- What is the difference among
--wait,--atomic, and--cleanup-on-fail? - Why are CRDs difficult to upgrade through Helm?
- What does
helm testprove and not prove? - How would you force a Pod rollout after a ConfigMap change?
- How would you test a chart across a matrix of optional features?
Flux and GitOps¶
- Trace a Git commit through Flux to a running Pod.
- Distinguish
GitRepository,OCIRepository, FluxKustomization, andHelmRelease. - Distinguish a Flux Kustomization from
kustomization.yaml. - What happens when Git is unavailable after an application is running?
- How do
interval, watched-resource events, webhook Receivers, and manual reconcile interact? - What does
prunedo, and when is it dangerous? - How do
wait,healthChecks, anddependsOndiffer? - How does
valuesFromprecedence work? - Describe HelmRelease install and upgrade remediation choices.
- How does Helm drift detection work, and when would you ignore a field?
- Why is a manual Helm rollback usually not durable under Flux?
- How would you structure infrastructure and applications across environments?
Security and platform¶
- Why is Workload Identity better than a service-account JSON key?
- Which identity pulls a private OCI chart in this design?
- How would you restrict what a Flux Kustomization or HelmRelease can create?
- Why is a base64 Kubernetes Secret unsafe to commit?
- Compare SOPS decryption with an external secret operator.
- How do you protect the software supply chain from Git to chart to image?
- What repository and branch controls matter for GitOps?
- What is the blast radius of source-controller compromise?
Troubleshooting and design¶
- An
OCIRepositoryis Ready butHelmReleaseis not. Where do you look? - Git shows the new commit but the cluster does not. Walk through diagnosis.
- A release is Ready but Pods are not. Which layer owns the failure?
- Flux repeatedly reverts an HPA's replica changes. How do you reason about it?
- A chart works locally but fails under Flux. List likely differences.
- How do you promote a chart from staging to production without mutable tags?
- When would CI push delivery be preferable to Flux?
- What would you monitor and alert on in a production Flux installation?
Evaluation rubric¶
A strong answer includes the control flow, owning component, observable evidence, security boundary, and at least one trade-off. A weak answer only defines a command or says to reinstall/reconcile without locating the failure.