agentscope-ai/agentscope · error · RuntimeError

Bubblewrap read_file failed

Error message

Bubblewrap read_file failed (exit {result.exit_code}): {result.stderr.decode(errors='replace')}

What it means

A catch-all RuntimeError from read_file when the sandboxed cat command fails with an unexpected exit code. The message embeds the exit code and the sandboxed command's stderr, which is the primary diagnostic.

Solutions

1. Read the embedded stderr in the exception message to identify the underlying cause
2. Verify bwrap works standalone (bwrap --ro-bind / / echo ok)
3. Check file permissions and mount health inside the sandbox
4. If bwrap setup fails, re-provision the workspace

null

null

try:
    data = await ws.read_file(p)
except RuntimeError as e:
    log.error('sandbox read failed: %s', e)  # stderr is embedded

When it happens

Trigger: Any failure of the inner `sh -c cat` that is neither exit 66 (not found) nor 65 (outside writable mounts): e.g. permission bits on the file inside the sandbox, I/O errors, or bwrap itself failing to launch.

Common situations: Read-only or corrupted mounts inside the sandbox, a broken bwrap installation, or disk errors. Inspect result.stderr embedded in the message.

Related errors

• Bubblewrap write failed
• host_cache_dir must not be a symbolic link.
• host_workdir must be a directory.
• must not overlap .
• not found in Bubblewrap sandbox

AI-assisted analysis of agentscope-ai/agentscope@e90f1c7592 (2026-08-28). Data as JSON: /api/errors/bca7f09f482d8b53. Report an issue: GitHub.

Appendix: source

Thrown at src/agentscope/workspace/_bubblewrap/_bubblewrap_backend.py:276

                    'resolved=$(realpath -e -- "$1") || exit 66; '
                    'case "$resolved" in '
                    "/workspace|/workspace/*|/tmp|/tmp/*) "
                    'cat -- "$resolved" ;; '
                    "*) exit 65 ;; "
                    "esac"
                ),
                "sh",
                sandbox_path,
            ],
        )
        if result.exit_code == 66:
            raise FileNotFoundError(f"not found in Bubblewrap sandbox: {path}")
        if result.exit_code == 65:
            raise PermissionError(
                f"path escapes writable Bubblewrap mounts: {path}",
            )
        if not result.ok():
            raise RuntimeError(
                "Bubblewrap read_file failed "
                f"(exit {result.exit_code}): "
                f"{result.stderr.decode(errors='replace')}",
            )
        return result.stdout

    async def write_file(self, path: str, data: bytes) -> None:
        """Write raw bytes, refusing every symbolic-link final component."""
        await self._write_via_cat(path, data)

    async def write_stream(
        self,
        path: str,
        stream: AsyncIterator[bytes],
    ) -> None:
        """Stream bytes into ``path``, chunk by chunk through ``cat``."""
        await self._write_via_cat(path, stream)

View on GitHub (pinned to e90f1c7592)