What causes "Failed to read %s" in hashicorp/terraform?

For stdin, ensure the producer process completes before terraform fmt reads — write to a temp file first, then `terraform fmt - < tmp.tf`.

hashicorp/terraform · warning

Failed to read %s

Error message

Failed to read %s

What it means

Returned by FmtCommand.processFile (fmt.go:179) when io.ReadAll(r) fails while reading the content of a file or stdin stream slated for formatting. processFile is called for both stdin ('<stdin>') and on-disk files; the path label distinguishes them. This is a low-level I/O failure after the source was already opened.

Source

Thrown at internal/command/fmt.go:179

			if !fmtd {
				diags = diags.Append(fmt.Errorf("Only .tf, .tfvars, and .tftest.hcl files can be processed with terraform fmt"))
				continue
			}
		}
	}

	return diags
}

func (c *FmtCommand) processFile(path string, r io.Reader, w io.Writer, isStdout bool) tfdiags.Diagnostics {
	var diags tfdiags.Diagnostics

	log.Printf("[TRACE] terraform fmt: Formatting %s", path)

	src, err := io.ReadAll(r)
	if err != nil {
		diags = diags.Append(fmt.Errorf("Failed to read %s", path))
		return diags
	}

	// Register this path as a synthetic configuration source, so that any
	// diagnostic errors can include the source code snippet
	c.registerSynthConfigSource(path, src)

	// File must be parseable as HCL native syntax before we'll try to format
	// it. If not, the formatter is likely to make drastic changes that would
	// be hard for the user to undo.
	_, syntaxDiags := hclsyntax.ParseConfig(src, path, hcl.Pos{Line: 1, Column: 1})
	if syntaxDiags.HasErrors() {
		diags = diags.Append(syntaxDiags)
		return diags
	}

	result := c.formatSourceCode(src, path)

View on GitHub (pinned to c9def3e214)

Solutions

For stdin, ensure the producer process completes before terraform fmt reads — write to a temp file first, then `terraform fmt - < tmp.tf`.
For on-disk files, retry the read; if it persists, run `fsck`/check disk health and restore from version control.
If on a network filesystem, copy the file to local storage and fmt the copy.
Close nothing prematurely: pipe with `cat file | terraform fmt -` only if `cat` is reliable; prefer file paths.

Example fix

# before
$ flaky-gen | terraform fmt -
Failed to read <stdin>   # producer died, pipe closed

# after
$ flaky-gen > tmp.tf || { echo 'producer failed'; exit 1; }
$ terraform fmt - < tmp.tf

Defensive patterns

Strategy: try-catch

Validate before calling

// Read stdin into a buffer first so I/O errors surface before fmt runs
buf, err := io.ReadAll(os.Stdin)
if err != nil {
    return fmt.Errorf("failed to read stdin: %w", err)
}
if _, err := fcmd.fmt([]string{}, bytes.NewReader(buf), out); err != nil { return err }

Try / catch

// For file inputs, retry once on a transient read error; for stdin, fail fast with a clear message.
if _, err := io.ReadAll(r); err != nil {
    if path != "<stdin>" && isTransient(err) {
        time.Sleep(100 * time.Millisecond)
        // reopen and retry once
    } else {
        return fmt.Errorf("%s unreadable: %w", path, err)
    }
}

Prevention

Materialize piped input into a temp file before formatting so a producer crash is detectable up front.
Keep .tf files on local, healthy storage.
In CI, verify the producer's exit code before piping to fmt.
Avoid network filesystems for files being formatted.

When it happens

Trigger: io.ReadAll returns an error: a read error on the underlying reader (disk I/O error, network filesystem read failure for an opened file, stdin pipe broken because the producer process exited early, or a file reader that hits a bad block).

Common situations: Piping into fmt from a process that crashes (`broken pipe`): `terraform fmt - < <(flaky-cmd)`; reading a .tf file on a flaky NFS share; stdin closed prematurely by a CI step; disk hardware error on a workstation; a file truncated by another process mid-read.

Related errors

AI-assisted analysis of hashicorp/terraform@c9def3e214 (2026-08-07). Data as JSON: /api/errors/15359d5cef988ace. Report an issue: GitHub.