router-for-me/CLIProxyAPI · error

failed to write token to file: %w

Error message

failed to write token to file: %w

What it means

json.NewEncoder(f).Encode(data) failed writing the serialized token to the opened file — an I/O error during the actual write or flush, not during open. Common causes: disk full (ENOSPC), quota exceeded, or the filesystem being remounted read-only after open. Because the encoder streams directly to the file, a mid-write failure can leave a truncated (invalid) JSON file behind.

Source

Thrown at internal/auth/codex/token.go:81

	// Merge metadata using helper
	data, errMerge := misc.MergeMetadata(ts, ts.Metadata)
	if errMerge != nil {
		return fmt.Errorf("failed to merge metadata: %w", errMerge)
	}

	f, err := os.Create(authFilePath)
	if err != nil {
		return fmt.Errorf("failed to create token file: %w", err)
	}
	defer func() {
		if errClose := f.Close(); errClose != nil {
			log.Errorf("codex token storage: close token file error: %v", errClose)
		}
	}()

	if err = json.NewEncoder(f).Encode(data); err != nil {
		return fmt.Errorf("failed to write token to file: %w", err)
	}
	return nil
}

View on GitHub (pinned to 78f0c4079e)

Solutions

1. Check space: `df -h <auth-dir>` and inodes with `df -i <auth-dir>`; free space if full.
2. After fixing, verify the auth file is valid JSON (a truncated write leaves a corrupt file) and re-login/refresh if not.
3. Raise disk quota / enlarge the container volume.
4. Alert on disk usage for the volume hosting auths/ so credential saves never hit ENOSPC.
5. Consider a storage backend (Postgres/object store) where writes are transactional if local-disk truncation is a recurring risk.

Example fix

# before: auths on a full rootfs
df -h /  # 100%
# after: free space or move auth-dir to a healthy volume
cli-proxy-api --config config.yaml   # with auth-dir: /data/auths

// Check writable space before persisting credentials
var stat syscall.Statfs_t
if err := syscall.Statfs(filepath.Dir(path), &stat); err == nil {
    free := stat.Bavail * uint64(stat.Bsize)
    if free < 1<<20 { // < 1 MiB
        return errors.New("auth volume nearly full; token save would be truncated")
    }
}

if err := ts.SaveTokenToFile(path); err != nil {
    if strings.Contains(err.Error(), "failed to write token to file") {
        // ENOSPC/quota: free space, then re-save; verify file is valid JSON afterwards
    }
}

When it happens

Trigger: Disk full at the moment of saving credentials; per-user quota hit on the volume; file opened then the underlying device/vanished (ejected volume, dead NFS server); fcntl locks not a factor but VM disk limits are.

Common situations: Small container filesystems filling up with logs; long-running servers whose volume hits quota right when a refreshed token is persisted; network storage blipping mid-write.

Related errors

• failed to create directory: %v
• failed to create token file: %w
• failed to save refreshed auth: %w
• failed to read auth file: %w
• failed to create directory: %v

AI-assisted analysis of router-for-me/CLIProxyAPI@78f0c4079e (2026-08-15). Data as JSON: /api/errors/ac98952f275c7462. Report an issue: GitHub.