agentscope-ai/agentscope · error · FileNotFoundError

not found in Pod: {path}

Error message

not found in Pod: {path}

What it means

Raised by K8sBackend.read_file() when `tar cf - -C <dir> <basename>` inside the Pod fails and its stderr contains 'No such file' or 'not found'. This is the explicit missing-file branch: the requested path does not exist inside the Kubernetes Pod. Note this backend reads files by tar-streaming them out of the container, so tar's error text is the signal.

Source

Thrown at src/agentscope/workspace/_k8s/_k8s_backend.py:235

        Returns:
            `bytes`:
                The raw file contents.

        Raises:
            `FileNotFoundError`:
                If the path does not exist inside the Pod.
        """
        dirname = posixpath.dirname(path) or "/"
        basename = posixpath.basename(path)

        result = await self.exec_shell(
            ["tar", "cf", "-", "-C", dirname, basename],
            cwd="/",
        )
        if not result.ok():
            stderr_text = result.stderr.decode(errors="replace")
            if "No such file" in stderr_text or "not found" in stderr_text:
                raise FileNotFoundError(
                    f"not found in Pod: {path}",
                )
            raise FileNotFoundError(
                f"not found in Pod: {path} (tar stderr: {stderr_text})",
            )

        try:
            tar = tarfile.open(fileobj=io.BytesIO(result.stdout), mode="r")
        except tarfile.ReadError as exc:
            raise FileNotFoundError(
                f"not found in Pod: {path}",
            ) from exc

        try:
            for member in tar.getmembers():
                if member.isfile():
                    f = tar.extractfile(member)
                    if f:

View on GitHub (pinned to e90f1c7592)

Solutions

1. Verify the path exists first: `await backend.exec_shell(["test", "-f", path])` and check exit code
2. Fix path construction to use the in-Pod absolute path
3. Catch FileNotFoundError and regenerate/recreate the file if it's a transient spill artifact
4. Serialize read+delete around shared tempfiles to avoid races

Example fix

// before
data = await backend.read_file(spill_path)  # FileNotFoundError
// after
if (await backend.exec_shell(["test", "-f", spill_path])).ok():
    data = await backend.read_file(spill_path)
else:
    data = b""

if (await backend.exec_shell(["test", "-f", path])).ok():
    data = await backend.read_file(path)
else:
    data = None

try:
    data = await backend.read_file(path)
except FileNotFoundError:
    data = None  # treat missing as empty

When it happens

Trigger: `await backend.read_file(path)` where path doesn't exist in the Pod — e.g. a tempfile already cleaned up, a gateway body_file spilled and deleted, or a wrong absolute path.

Common situations: Reading a gateway spill file (body_file) that another request already deleted; racing cleanup code; paths built for a different container or the host rather than the Pod; file removed between existence check and read.

Related errors

• not found in Pod: {path} (tar stderr: {stderr_text})
• write to {path!r} failed: {command[0]} exited {exit_code}: {
• {entry.path!r} is larger than its declared {entry.size} byte
• {entry.path!r} declared {entry.size} bytes but sent {written
• not found in container: {path}\nstderr: {result.stderr.decod

AI-assisted analysis of agentscope-ai/agentscope@e90f1c7592 (2026-08-28). Data as JSON: /api/errors/48adda2e2b2fed69. Report an issue: GitHub.