kubernetes/kops · error

creating file %q: %w

Error message

creating file %q: %w

What it means

After ensuring the parent directory exists, dumpGVRNamespaces creates the output file for a resource job via os.Create(resPath). If creation fails, the OS error is wrapped as "creating file %q: %w". Like the directory error, this is a local write failure, not from the Kubernetes API.

Source

Thrown at pkg/dump/resourcedumper.go:216

				continue
			}
			results <- resourceDumpResult{
				err: fmt.Errorf("listing resources for %v: %w", job, err),
			}
			continue
		}
		resPath := path.Join(d.artifactsDir, "cluster-info", fmt.Sprintf("%v.%v", job.String(), d.output))
		err = os.MkdirAll(path.Dir(resPath), 0755)
		if err != nil {
			results <- resourceDumpResult{
				err: fmt.Errorf("creating directory %q: %w", resPath, err),
			}
			continue
		}
		resFile, err := os.Create(resPath)
		if err != nil {
			results <- resourceDumpResult{
				err: fmt.Errorf("creating file %q: %w", resPath, err),
			}
			continue
		}

		err = resourceList.EachListItem(func(obj runtime.Object) error {
			o, err := meta.Accessor(obj)
			if err != nil {
				return err
			}
			o.SetManagedFields(nil)
			if err := maskObject(obj); err != nil {
				return err
			}
			return nil
		})
		if err != nil {
			results <- resourceDumpResult{
				err: fmt.Errorf("creating accessor for %v: %w", job, err),

View on GitHub (pinned to 4c8573c808)

Solutions

1. Fix ownership/permissions of the artifacts/cluster-info directory so the running user can write.
2. Shorten --artifacts-dir or the output path to keep filenames under the filesystem limit.
3. Ensure no directory exists at the target file path; remove the collision.
4. Free disk space or move the dump to a volume with room.
5. If writing inside a container, mount a writable volume at the artifacts dir.

Example fix

// before
err = os.Create("/mnt/ro/cluster-info/pods.default.yaml") // permission denied
// after
chmod u+w /mnt/out/cluster-info  # or use a writable --artifacts-dir

if len(path.Base(resPath)) > 255 {
  return fmt.Errorf("output filename %q exceeds filesystem name limit", resPath)
}
if f, err := os.OpenFile(resPath, os.O_CREATE|os.O_WRONLY, 0o644); err != nil {
  return fmt.Errorf("cannot write to %s: %w", resPath, err)
} else {
  f.Close()
  os.Remove(resPath)
}

if err := run(); err != nil {
  var pe *os.PathError
  if errors.As(err, &pe) {
    switch {
    case errors.Is(pe.Err, syscall.ENAMETOOLONG): // shorten --artifacts-dir
    case errors.Is(pe.Err, syscall.EACCES):      // fix permissions
    case errors.Is(pe.Err, syscall.ENOSPC):      // free disk space
    }
  }
}

When it happens

Trigger: os.Create(<artifactsDir>/cluster-info/<gvr-namespace>.<output>) fails because the directory is not writable, the target name exists as a directory, the filename exceeds the filesystem's NAME_MAX, or the disk is full.

Common situations: Artifacts dir on a read-only volume; running kops as a non-root user against a root-owned dir; long resource/namespace names pushing the filename over 255 bytes on ext4; an output filename colliding with an existing directory; out-of-disk during a large cluster dump.

Related errors

• creating directory %q: %w
• unable to read snippet: %s, error: %s
• unable to read template: %s, error: %s
• encoding resources for %v: %w
• error creating output directory %q: %v

AI-assisted analysis of kubernetes/kops@4c8573c808 (2026-09-05). Data as JSON: /api/errors/601edcf79998262e. Report an issue: GitHub.