70+ FluxCD Interview Questions and Answers (2027)
72 FluxCD interview questions on CRD fields, defaults and flux CLI commands: GitRepository, Kustomization, HelmRelease, image automation, alerts.
FluxCD interview questions test whether you know Flux at the level of its custom resources and CLI: the fields on GitRepository, Kustomization, HelmRelease, ImagePolicy and Alert objects, what their defaults are, and which flux command answers which question. This page has 72 numbered questions with short, direct answers, grouped by experience level and followed by scenarios.
Flux is a set of Kubernetes controllers that pull desired state from Git, OCI registries or buckets and apply it to a cluster. Every answer below was checked against the Flux documentation at fluxcd.io in November 2026, when the docs listed v2.10 as the newest release line. If you want a broader first pass on GitOps concepts, start with our top FluxCD GitOps interview questions, which is the main page for this topic. This page goes one level deeper, into fields, defaults and commands.
Key takeaways
- Know the six controller areas and their API groups: source (
source.toolkit.fluxcd.io/v1), kustomize (v1), helm (v2), notification (v1) and the two image automation APIs (image.toolkit.fluxcd.io/v1). - Interviewers like defaults: a GitRepository with no
reftracksmaster, HelmReleasemaxHistorydefaults to 5, KustomizationdeletionPolicydefaults toMirrorPrune. - Field precedence questions come up often, for example
commitbeatstagbeatsbranchin a GitRepository ref, and inlinevaluesbeatvaluesFrom. - The debugging commands that separate juniors from seniors are
flux trace,flux tree,flux events,flux debugandflux diff kustomization. flux migratemust be run before a Flux minor version upgrade, according to the Flux CLI docs.- For experienced roles, expect scenarios where you have to name the exact field you would change.
How this page fits with the other Flux guides
| Page | Best for |
|---|---|
| Top FluxCD GitOps interview questions | Concepts, Flux vs Argo CD, repository layout, multi-tenancy |
| This page (70+ questions) | Field-level knowledge of every Flux CRD, defaults and CLI commands |
| Scenario-based FluxCD interview questions | Longer incident walk-throughs |
| FluxCD interview preparation guide | A study plan and lab order |
Flux custom resources at a glance
| Controller | Kinds | API version |
|---|---|---|
| source-controller | GitRepository, OCIRepository, HelmRepository, HelmChart, Bucket, ExternalArtifact, ArtifactGenerator | source.toolkit.fluxcd.io/v1 |
| kustomize-controller | Kustomization | kustomize.toolkit.fluxcd.io/v1 |
| helm-controller | HelmRelease | helm.toolkit.fluxcd.io/v2 |
| notification-controller | Provider, Alert, Receiver | notification.toolkit.fluxcd.io/v1 |
| image-reflector-controller | ImageRepository, ImagePolicy | image.toolkit.fluxcd.io/v1 |
| image-automation-controller | ImageUpdateAutomation | image.toolkit.fluxcd.io/v1 |
Questions for freshers (1 to 24)
1. Which API group and version does a GitRepository use today?
source.toolkit.fluxcd.io/v1. Older tutorials show v1beta2; if you see that in a repository, it is a sign the manifests need migrating.
2. What does Flux track if a GitRepository has no ref at all?
The master branch. Many repositories now use main, so leaving out ref is a common reason a new source never becomes Ready.
3. Which fields can you set under spec.ref?
branch, tag, semver, name and commit. You can set more than one, and later fields in that list win, so commit overrides the rest.
4. What keys does the secret for HTTPS basic auth need?
username and password. For a bearer token the key is bearerToken, and for a custom CA it is ca.crt.
5. What keys does an SSH secret need?
identity (the private key) and known_hosts. Add a password key only if the private key is passphrase protected.
6. What is the default value of spec.provider on a GitRepository?
generic, which means Flux uses the secret named in secretRef. The other values, aws, azure and github, use the cloud or GitHub App identity instead of a static secret.
7. What does a Kustomization's spec.path point to?
A directory inside the source artifact. If that directory has a kustomization.yaml, Flux builds it; if not, Flux generates one that includes every manifest it finds.
8. Is spec.prune optional on a Kustomization?
No. It is a required boolean in the v1 API. Setting it to true lets Flux garbage collect objects that were removed from Git.
9. Which source kinds can a Kustomization reference?
GitRepository, OCIRepository, Bucket and ExternalArtifact. ExternalArtifact needs the ExternalArtifact=true feature gate.
10. What does spec.interval control?
How often the controller reconciles the object even when nothing changed. A change to the source artifact triggers reconciliation without waiting for the interval.
11. What does spec.retryInterval do and what is its default?
It sets how soon Flux retries after a failed reconciliation. If you do not set it, it uses the value of spec.interval.
12. Which command checks that a cluster meets Flux prerequisites?
flux check --pre before installing, and flux check afterwards to confirm the controllers and CRDs are healthy.
13. What is the difference between flux install and flux bootstrap?
flux install only deploys the controllers. flux bootstrap also commits the Flux manifests to a Git repository and configures Flux to manage itself from that repository.
14. Where does flux bootstrap github read the token from?
The GITHUB_TOKEN environment variable. If it is not set, the command prompts for it or reads it from standard input.
15. If the target repository does not exist, what does bootstrap do?
It creates it, and the new repository is private unless you pass --private=false.
export GITHUB_TOKEN=<your-token>
flux bootstrap github \
--token-auth \
--owner=my-github-username \
--repository=my-repository-name \
--branch=main \
--path=clusters/my-cluster \
--personal
16. What does --personal mean on bootstrap?
The owner is a user account, not an organization. Leave it out when bootstrapping into an organization repository.
17. Which command lists every Flux object and its readiness?
flux get all -A. Narrower forms include flux get sources git -A and flux get kustomizations -A.
18. What does flux reconcile kustomization NAME --with-source do?
It asks Flux to fetch the source first and then reconcile the Kustomization immediately, instead of waiting for both intervals.
19. How do you pause one object?
flux suspend kustomization NAME sets spec.suspend: true. flux resume kustomization NAME sets it back and triggers a reconcile.
20. Which command shows the Kubernetes events for Flux objects?
flux events, optionally with --for Kustomization/NAME to focus on one object.
21. Which command shows controller logs filtered for one object?
flux logs --kind=Kustomization --name=NAME. Add --level=error to see only errors.
22. What does flux export produce?
Clean YAML for existing Flux objects, without status fields. Engineers use it to move objects created by the CLI into Git.
23. What does flux uninstall remove?
The Flux controllers, CRDs and namespace. It leaves the workloads Flux deployed in place, which is what you usually want during a migration.
24. Does Flux ship a web UI?
No. The Flux project points to UIs built by the ecosystem; the core project is CLI and API driven.
Questions for 2 to 5 years of experience (25 to 50)
25. What does spec.timeout on a Kustomization cover?
The build, apply and health check phases. If health checks do not pass inside the timeout, the Kustomization reports Ready False.
26. What is the difference between spec.wait and spec.healthChecks?
wait: true makes Flux wait for every applied object to become ready. healthChecks lists specific objects to watch. When wait is true, healthChecks is ignored.
27. What are healthCheckExprs?
CEL expressions that tell Flux how to judge the health of a custom resource it does not understand. Each entry needs an apiVersion and a current expression, with optional failure and in-progress expressions.
28. What does dependsOn with readyExpr add?
A CEL expression that replaces the default "is Ready" test for the dependency. You use it when "ready" for your purposes means something more specific, such as a status field reaching a value.
29. What is the difference between postBuild.substitute and substituteFrom?
substitute is an inline map of variables. substituteFrom reads variables from ConfigMaps or Secrets; each entry has an optional flag that defaults to false, so a missing ConfigMap fails the build.
30. What does substituteStrategy control?
Whether substitution runs only on objects that contain variables (WithVariables, the default) or on every object (Always).
31. How do you debug which variables Flux substituted?
Run flux debug kustomization NAME --show-vars. The command prints values from referenced Secrets too, so do not paste its output into tickets.
32. What does targetNamespace do, and what is the catch?
It sets the namespace on every namespaced object the Kustomization applies. The namespace must already exist or be defined in the same Kustomization.
33. When would you set spec.force: true?
When an apply fails because a field is immutable, for example a Job template or a Service clusterIP. Flux then deletes and recreates the object, so use it only where that is safe.
34. Name the four deletionPolicy values.
MirrorPrune (default, follows the prune setting), Delete, WaitForTermination and Orphan. The policy decides what happens to applied objects when the Kustomization itself is deleted.
35. How do you decrypt SOPS-encrypted secrets with a Kustomization?
Set spec.decryption.provider: sops and point secretRef at the secret holding the age or PGP key, or use serviceAccountName for cloud KMS identity. sops is the only supported provider.
apiVersion: kustomize.toolkit.fluxcd.io/v1
kind: Kustomization
metadata:
name: apps
namespace: flux-system
spec:
interval: 10m
sourceRef:
kind: GitRepository
name: flux-system
path: ./apps/production
prune: true
timeout: 2m
decryption:
provider: sops
secretRef:
name: sops-age
36. What does serviceAccountName on a Kustomization do?
Flux impersonates that service account when applying, so the Kustomization can only touch what the account's RBAC allows. This is the basis of Flux multi-tenancy.
37. What is the difference between HelmRelease spec.chart and spec.chartRef?
chart is a template from which Flux creates a HelmChart source. chartRef points at an existing OCIRepository or HelmChart. You set exactly one of them.
38. What do install.remediation.retries and upgrade.remediation.retries default to?
Zero, so a failed install or upgrade is not retried unless you set them. A negative value means retry forever.
39. What does remediateLastFailure default to?
false for installs. For upgrades it defaults to true when at least one retry is configured, which means the last failed upgrade gets rolled back.
40. What are the two remediation strategies?
RemediateOnFailure rolls back or uninstalls and then retries. RetryOnFailure retries without remediating, waiting retryInterval (default 5 minutes) between attempts.
41. In which order are Helm values merged?
valuesFrom entries merge in the order listed, later ones overriding earlier ones, and inline spec.values override all of them.
42. What are the two drift detection modes?
warn reports drift as an event. enabled also corrects it. Use the ignore list with JSON Pointer paths to skip fields another controller owns, such as replicas managed by an HPA.
43. What is the default HelmRelease maxHistory?
5 release revisions. Raise it if you need longer rollback history; lower it if Helm release secrets are bloating etcd.
44. What release name does Flux use if you do not set one?
[<target namespace>-]<name>, that is, the HelmRelease name prefixed with the target namespace when one is set.
45. How do you run Helm tests after an upgrade?
Set spec.test.enable: true. ignoreFailures and filters let you decide which test failures block the release.
46. What are post renderers for?
Applying Kustomize patches and images overrides after Helm renders the chart, for changes the chart's values do not expose.
47. How do you see the final merged values Flux passed to Helm?
flux debug helmrelease NAME --show-values. Like the Kustomization variant, it can print secret values.
48. Write the marker comment that lets image automation update a Deployment image.
Append a setter comment that names the policy as namespace:name.
image: ghcr.io/my-org/my-app:4.0.6 # {"$imagepolicy": "flux-system:my-policy"}
49. How do you update only the tag in Helm values?
Add :tag to the policy reference: tag: 4.0.6 # {"$imagepolicy": "flux-system:my-policy:tag"}. There are also :name and :digest selectors.
50. What does the :digest setter require?
The ImagePolicy must set spec.digestReflectionPolicy to IfNotPresent or Always, otherwise there is no digest to write.
Questions for senior engineers (51 to 62)
51. How do you verify commit signatures before Flux applies anything?
Set spec.verify on the GitRepository with mode HEAD, Tag or TagAndHEAD, and a secretRef holding trusted keys (.asc for PGP, .sshpub for SSH). An unsigned or wrongly signed commit leaves the source not Ready.
52. A monorepo is slow to fetch. Which GitRepository fields help?
sparseCheckout limits the checkout to listed directories, and ignore excludes paths from the artifact. Both reduce artifact size and the work downstream Kustomizations do.
53. What does spec.include do on a GitRepository?
It copies the artifact of another GitRepository into this one, using fromPath and toPath. Teams use it to pull shared base manifests into an application source without submodules.
54. Why would you ship manifests as OCI artifacts instead of reading Git directly?
Registries scale and replicate better than Git servers, artifacts are immutable and can be signed with cosign, and clusters then need registry access only. CI pushes with flux push artifact and clusters pull through an OCIRepository.
flux push artifact oci://ghcr.io/org/config/app:$(git rev-parse --short HEAD) \
--path="./deploy" \
--source="$(git config --get remote.origin.url)" \
--revision="$(git branch --show-current)@sha1:$(git rev-parse HEAD)"
55. What do flux tag artifact and flux pull artifact add to that workflow?
tag adds a tag such as production to an already pushed digest, which is a clean promotion step. pull downloads an artifact locally so you can inspect exactly what a cluster received.
56. What is the ExternalArtifact kind for?
It lets a third-party controller produce an artifact that Flux Kustomizations can consume as a source. It sits behind the ExternalArtifact=true feature gate.
57. How should you upgrade Flux across a minor version?
Run flux migrate first to move stored objects to the newest API versions, and flux migrate -f . on the Git repository so manifests stop using removed fields. Then upgrade the controllers, normally by re-running bootstrap or committing new gotk-components.yaml.
58. How do you preview the effect of a change on a live cluster before merging?
flux diff kustomization NAME --path ./local/dir builds the local directory and diffs it against what is applied. In CI, combine it with flux build kustomization to fail pull requests that do not build.
59. How do you route only failures from production to an incident channel?
Create a Provider for the channel and an Alert with eventSeverity: error and eventSources listing the production Kustomizations and HelmReleases. With info (the default), every event is forwarded.
apiVersion: notification.toolkit.fluxcd.io/v1
kind: Alert
metadata:
name: prod-failures
namespace: flux-system
spec:
providerRef:
name: oncall-slack
eventSeverity: error
eventSources:
- kind: Kustomization
name: '*'
- kind: HelmRelease
name: '*'
60. How do you make Git pushes trigger reconciliation instantly?
Create a Receiver for your Git host, expose the notification-controller webhook endpoint through an Ingress, and register the generated URL as a webhook. The Receiver then requests reconciliation of the listed resources.
61. How do you onboard a tenant with least privilege?
flux create tenant generates a namespace, service account and RoleBinding. Tenant Kustomizations then set serviceAccountName so they cannot apply outside their namespace.
62. Which command gives you a quick count of Flux objects and their states across the cluster?
flux stats. It is a fast way to see how many objects are failing or suspended before you dig into individual ones.
Scenario questions (63 to 72)
63. A Deployment in production has a strange setting. How do you find which Flux object applied it?
Run flux trace -n apps deployment my-app. It walks back from the Deployment to the Kustomization or HelmRelease and the source revision that produced it.
64. You need to see everything one Kustomization manages before deleting it.
flux tree kustomization NAME lists the objects in its inventory, including nested Kustomizations and HelmReleases. Then decide on deletionPolicy before you remove it.
65. A new GitRepository never becomes Ready and the error says the reference was not found.
Check spec.ref. If it is missing, Flux looks for master, and a repository that only has main fails. Set ref.branch: main.
66. A Kustomization fails with "variable not found" after a ConfigMap was renamed.
A substituteFrom entry points to the old name and optional defaults to false. Fix the reference, or mark it optional if the variables really are optional, then check the result with flux debug kustomization --show-vars.
67. An upgrade of a HelmRelease failed and nothing was rolled back.
Upgrade remediation retries default to 0, so no remediation ran. Set upgrade.remediation.retries and confirm remediateLastFailure is true, then resume reconciliation.
68. HelmRelease drift correction keeps fighting the HPA over replicas.
Add an ignore rule under driftDetection for the replicas path of that Deployment, or switch to warn mode while you decide.
69. A Job manifest changed and the apply fails on an immutable field.
Set force: true on that Kustomization, or better, move Jobs into a separate Kustomization with force so the rest of the app is not recreated on errors.
70. After a Flux upgrade, some objects show warnings about deprecated API versions.
The stored objects or Git manifests still use old versions. Run flux migrate in the cluster and flux migrate -f . in the repository, commit the changes and reconcile.
71. Image automation pushes commits but the Deployment never changes.
Check the marker comment format and that the policy name in it matches namespace:name. Then confirm the Kustomization that applies the file is reconciling the commit image automation made, with flux get kustomizations.
72. Security asks for proof that only signed changes reach production.
Enable verify on the production GitRepository, or move to OCI artifacts signed with cosign and verified by the OCIRepository. Pair it with Alerts on verification failures so a bad commit is seen immediately.
Flux CLI commands for each interview question type
| Interview question | Command |
|---|---|
| Is the cluster ready for Flux? | flux check --pre |
| What is failing right now? | flux get all -A --status-selector ready=false |
| Who manages this object? | flux trace -n NS deployment NAME |
| What does this Kustomization own? | flux tree kustomization NAME |
| What will my change do? | flux diff kustomization NAME --path ./dir |
| Which variables were substituted? | flux debug kustomization NAME --show-vars |
| Which Helm values were used? | flux debug helmrelease NAME --show-values |
| Prepare for an upgrade | flux migrate and flux migrate -f . |
How to practise these questions
- Create a kind or minikube cluster and bootstrap Flux into a test repository.
- Write one GitRepository, one Kustomization with SOPS decryption and one HelmRelease with remediation retries from memory, then compare them with the docs.
- Break things on purpose: remove
ref, rename a ConfigMap used bysubstituteFrom, push a bad chart value. Use the commands in the table above to diagnose each one. - Read the Argo CD and Flux comparison so you can answer the "why Flux" question, and skim our Argo CD interview questions if the role uses both.
Flux sits on top of Kubernetes, so weak Kubernetes basics show up quickly in these interviews. Our CKA training in Pune (classroom in Kharadi or live online) covers the cluster skills these questions assume, and the 200+ DevOps interview questions guide links to every other tool page. For background on the model itself, see what GitOps is.
FAQ
What API version should Flux manifests use in 2027?
Use source.toolkit.fluxcd.io/v1, kustomize.toolkit.fluxcd.io/v1, helm.toolkit.fluxcd.io/v2, notification.toolkit.fluxcd.io/v1 and image.toolkit.fluxcd.io/v1. Run flux migrate -f . to update older manifests.
What branch does Flux use if no ref is set?
A GitRepository without spec.ref tracks master. Set ref.branch: main explicitly for repositories that use main.
What is the default HelmRelease maxHistory?
Five revisions. You can raise or lower it with spec.maxHistory.
Does a failed HelmRelease upgrade roll back by default?
No. Upgrade remediation retries default to zero. Set upgrade.remediation.retries to at least 1 to get a rollback of the last failure.
How do I find which Flux object created a Deployment?
Run flux trace -n NAMESPACE deployment NAME. It shows the Kustomization or HelmRelease and the source revision.
What must I run before a Flux minor upgrade?
The Flux CLI docs say flux migrate must be run before a minor version upgrade, so stored objects move to the latest API versions.
Is this page enough on its own for a Flux interview?
It covers field and command depth. Pair it with the top FluxCD GitOps questions page for concepts and the scenario page for long incident answers.
Official references
What's Your Reaction?
Like
0
Dislike
0
Love
0
Funny
0
Angry
0
Sad
0
Wow
0