Join our Newsletter — 33% off our NHI Course
Home› FAQ› Cyber Security› What is the difference between kubectl create and…
Cyber Security

What is the difference between kubectl create and kubectl apply for managing Kubernetes manifests?

← Back to all FAQ
By NHI Mgmt Group Editorial Team Updated October 31, 2026 Domain: Cyber Security

kubectl create is imperative and creates a new object from a file or command input, but it will fail if that resource already exists. kubectl apply is declarative and reconciles the live object with the manifest, creating it if needed or updating it when the specification changes. Use apply when you want repeatable configuration management.

How kubectl create and kubectl apply differ in day-to-day manifest management

kubectl create is best understood as a one-time object creation command. It takes the manifest you provide and sends a create request to the cluster, which means it is useful for first-time provisioning but not for managing later changes. kubectl apply treats the manifest as the desired state and uses it as a repeatable input for ongoing configuration management.

The practical difference is lifecycle behaviour. With create, the command succeeds only when the object does not already exist, so it is closer to a provisioning action than a reconciliation workflow. With apply, Kubernetes compares the manifest to the live object and updates fields to match, which is why it fits Git-driven or CI-driven deployment patterns better.

That distinction matters when manifests are reused across environments. A team using create must handle object existence, drift, and redeployments separately, while apply lets the manifest remain the source of truth for iterative changes such as replica counts, labels, environment variables, or resource limits.

Why declarative reconciliation is usually the safer operational default

Most Kubernetes teams want repeatability more than a single successful launch. Declarative workflows make it easier to review changes, compare desired state with live state, and re-run the same manifest without having to guess whether the object already exists. That is why apply aligns better with promotion between dev, staging, and production.

create still has a place when you want an explicit one-off action, such as bootstrapping a resource that will then be managed elsewhere or testing a manifest interactively. But once a workload is meant to be maintained over time, imperative creation becomes brittle because it does not express intent about updates, only initial instantiation.

If your operational model depends on repeatable deployments, the key judgement is whether you want the cluster to accept a new object or converge an existing one. apply answers the second question, which is usually the one that matters after the first rollout.

What trips teams up when switching between create and apply

The most common mistake is assuming the two commands are interchangeable. They are not. A manifest that works with apply may fail with create because the object already exists, and a workflow built around create often breaks when teams try to re-run it as part of automated delivery.

Another subtle issue is ownership of changes. apply is most effective when one pipeline or team is the authoritative writer for a resource. If multiple actors mutate the same object in different ways, reconciliation becomes harder to reason about and unexpected field changes can appear during future applies.

For that reason, teams should be clear about whether they are using manifests for provisioning, for ongoing reconciliation, or for both. Mixing those goals without a convention usually leads to drift, partial updates, or manual edits that are later overwritten.

Risk and Threat Considerations

Configuration drift is the main operational risk when manifests are not managed declaratively. If teams rely on ad hoc creation and manual follow-up changes, the live cluster can diverge from the intended configuration, which increases the chance of hidden privilege, exposure, or availability issues.

Failure mechanism: Imperative creation does not continuously reconcile state, so later edits, emergency fixes, or parallel manual changes can leave objects running with settings that are no longer reflected in source control.

Impact: Drift can make deployments harder to audit, harder to reproduce, and easier to misconfigure, especially when the same manifest is promoted across environments or reused by automation.

Standards & Framework Alignment

This section maps relevant standards and security frameworks to the operational risks and controls described in this guidance.

NIST CSF 2.0 and NIST SP 800-53 Rev 5 set the governance and control requirements practitioners need to meet.

FrameworkControl / ReferenceRelevance
NIST CSF 2.0PR.PS-02 — Software, Firmware and Information IntegrityDeclarative manifests help keep deployed configuration aligned to the approved source state.
Recommendation — Use PR.PS-02 to keep cluster configuration changes traceable and aligned to approved desired state.
NIST SP 800-53 Rev 5CM-2 — Baseline Configurationkubectl apply supports maintaining an approved configuration baseline for Kubernetes objects.
CM-6 — Configuration SettingsThe choice between create and apply affects how configuration settings are established and preserved.
Recommendation — Maintain approved Kubernetes baselines and reapply them to keep live state consistent. Define and enforce configuration settings through a repeatable deployment process.

Practitioner Guidance

What to prioritise: Use apply for resources that should remain aligned to a manifest over time, and reserve create for one-time bootstrapping or deliberately ephemeral objects. That choice should be explicit in your deployment process, not left to operator preference.

What to verify: Confirm that your pipeline has one clear source of truth for each resource, and check whether the resource is expected to be recreated, patched, or continuously reconciled. If humans and automation both touch the same object, document which path wins.

Practitioner takeaway: If the question is “what should the cluster look like after the next change?”, apply is usually the right tool; if the question is “does this object already exist?”, create is the narrower, imperative choice.

Free weekly newsletter

Subscribe to the NHI & AI Identity Journal

The latest on NHI and Agentic AI security – articles, research, breaches, news and events every week.

Bonus 33% off our NHI Course when you subscribe.

NHIMG Editorial Note
Reviewed and updated by the NHIMG editorial team on October 31, 2026.
NHI Mgmt Group — the #1 independent authority on Non-Human Identity, IAM, and Agentic AI security. nhimg.org