kubectl "Error from server (NotFound)" on deploy in CI

By Kaveh Alemi·Latchkey

NotFound means the API server has no object matching what you asked for. In CI this is usually a namespace that was never created, a typo in a name, or operating on a resource that was already deleted.

What this error means

A kubectl get/apply/rollout step returns "Error from server (NotFound)", naming the missing object or namespace.

Error from server (NotFound): namespaces "staging" not found

# or:
Error from server (NotFound): deployments.apps "api" not found

Common causes

Namespace not created

The deploy targets a namespace that the pipeline never created.

Wrong resource name

A typo in the deployment/service name does not match any existing object.

Acting on an already-deleted resource

A rollout status or delete runs against something a prior step removed.

How to fix it

Create the namespace before applying

kubectl create namespace staging --dry-run=client -o yaml | kubectl apply -f -

Verify names and existence

1. List the resources to confirm the exact name (kubectl get deploy -n <ns>).
2. Ensure prerequisite objects are applied before commands that reference them.
3. Order the pipeline so creation precedes status/rollout checks.

How to prevent it

• Create namespaces idempotently at the start of the deploy.
• Reference resources by names that exist in the manifests applied first.
• Use --ignore-not-found for cleanup steps that may run on absent objects.

Frequently asked questions

Related guides

References

• Kubernetes documentation
• GitHub Actions documentation