GoogleContainerTools/skaffold · error

kubectl create: %w

Error message

kubectl create: %w

What it means

The kubectl deployer uses `kubectl create --dry-run=client -o yaml` (ReadManifests) to materialize manifests with defaults (including namespace) before deploying. If the create command fails — including on a retry without the --namespace flag — this readManifestErr is returned. Usually the failure is a bad kubeconfig, unreachable cluster for validation, or unsupported flag combination.

Source

Thrown at pkg/skaffold/deploy/kubectl/cli.go:227

		dryRun += "=client"
	}

	args := c.args([]string{dryRun, "-oyaml"}, list...)

	if c.Flags.DisableValidation {
		args = append(args, "--validate=false")
	}
	argsWithoutNamespace := args
	if ns := util.ParseNamespaceFromFlags(c.Flags.Apply); ns != "" {
		args = append(args, "-n", ns)
	}

	buf, err := c.RunOut(ctx, "create", args...)
	if err != nil {
		// try running again without the `--namespace` flag
		buf, err = c.RunOutWithoutNamespace(ctx, "create", argsWithoutNamespace...)
		if err != nil {
			return nil, readManifestErr(fmt.Errorf("kubectl create: %w", err))
		}
	}

	var manifestList manifest.ManifestList
	manifestList.Append(buf)

	return manifestList, nil
}

func (c *CLI) args(commandFlags []string, additionalArgs ...string) []string {
	args := make([]string, 0, len(c.Flags.Global)+len(commandFlags)+len(additionalArgs))

	args = append(args, c.Flags.Global...)
	args = append(args, commandFlags...)
	args = append(args, additionalArgs...)

	return args
}

View on GitHub (pinned to a1189de023)

Solutions

1. Read the wrapped root cause; run the equivalent `kubectl create --dry-run=client -o yaml -f <manifest>` manually to see the full message.
2. Upgrade kubectl to a recent version matching your cluster (client dry-run features).
3. Fix kubeconfig: ensure the current context has a valid namespace or set one explicitly.
4. Verify all manifest paths referenced by the deploy config exist.

Example fix

// before
$ skaffold deploy  # kubectl create fails: no namespace in context
// after
kubectl config set-context --current --namespace=default
$ skaffold deploy

// Ensure context and namespace resolve before reading manifests
ctxName, _ := exec.Command("kubectl", "config", "current-context").Output()
ns, _ := exec.Command("kubectl", "config", "view", "--minify", "-ojsonpath={..namespace}").Output()
if strings.TrimSpace(string(ns)) == "" {
    return fmt.Errorf("context %s has no default namespace", ctxName)
}

if _, err := cli.ReadManifests(ctx, out, files, ns); err != nil {
    if strings.Contains(err.Error(), "kubectl create") {
        log.Errorf("manifest read failed: %v — verify kubeconfig and kubectl version", err)
    }
    return err
}

When it happens

Trigger: Both `c.RunOut(create args...)` and `c.RunOutWithoutNamespace(create argsWithoutNamespace...)` fail — e.g. kubectl can't resolve the current context/namespace, the --namespace flag is rejected by an old kubectl, or a referenced file (kustomize/-f path) is missing.

Common situations: Current kubeconfig context has no default namespace and dry-run needs it; very old kubectl binaries lacking client dry-run flags; malformed manifest files passed via skaffold kubectl deployer config.

Related errors

• nothing to deploy
• CONFIG_MISSING_MANIFEST_FILE_ERR
• rendering manifests: %w
• rendering manifests: %w
• could not read the hydrated manifest from %v: %w

AI-assisted analysis of GoogleContainerTools/skaffold@a1189de023 (2026-09-05). Data as JSON: /api/errors/9aedf9e31bd46126. Report an issue: GitHub.