infiniflow/ragflow · error · RuntimeError

Local execution output exceeded {self.max_output_bytes} byte

Error message

Local execution output exceeded {self.max_output_bytes} bytes.

What it means

Raised by LocalProvider._validate_output_size() after a child process finishes, when the combined UTF-8 byte length of captured stdout and stderr exceeds max_output_bytes (default 1 MiB). It fires after process completion (not during), so the process already ran to completion or its own timeout; the whole execution result is discarded and RuntimeError propagates from execute_code().

Source

Thrown at agent/sandbox/providers/local.py:314

        import resource

        self._set_resource_limit(resource.RLIMIT_CPU, self.timeout + 1)
        self._set_resource_limit(resource.RLIMIT_AS, self.max_memory_mb * 1024 * 1024)
        self._set_resource_limit(resource.RLIMIT_FSIZE, self.max_artifact_bytes)
        self._set_resource_limit(resource.RLIMIT_NOFILE, 64)

    @staticmethod
    def _set_resource_limit(kind: int, value: int) -> None:
        import resource

        _, hard = resource.getrlimit(kind)
        limit = value if hard == resource.RLIM_INFINITY else min(value, hard)
        resource.setrlimit(kind, (limit, limit))

    def _validate_output_size(self, stdout: str, stderr: str) -> None:
        output_size = len((stdout or "").encode("utf-8")) + len((stderr or "").encode("utf-8"))
        if output_size > self.max_output_bytes:
            raise RuntimeError(f"Local execution output exceeded {self.max_output_bytes} bytes.")

    def _collect_artifacts(self, artifacts_dir: Path) -> list[dict[str, Any]]:
        artifacts: list[dict[str, Any]] = []
        for path in sorted(artifacts_dir.rglob("*")):
            if path.is_symlink():
                raise RuntimeError(f"Artifact symlinks are not allowed: {path.name}")
            if path.is_dir():
                continue
            if not path.is_file():
                raise RuntimeError(f"Unsupported artifact entry: {path.name}")

            if len(artifacts) >= self.max_artifacts:
                raise RuntimeError(f"Local execution produced more than {self.max_artifacts} artifacts.")

            size = path.stat().st_size
            if size > self.max_artifact_bytes:
                raise RuntimeError(f"Artifact exceeds {self.max_artifact_bytes} bytes: {path.name}")

View on GitHub (pinned to 554fb1133a)

Solutions

Make the executed code write large outputs to files in the artifacts directory instead of stdout (files are governed by artifact limits, not output limits).
Increase 'max_output_bytes' at initialize() (schema max 10485760) if large stdout is legitimate.
Silence noisy stderr in the child: lower log levels, disable progress bars (e.g. TQDM_DISABLE=1, TRANSFORMERS_VERBOSITY=error), redirect pip output.
Return only a summary/pointer (path, row count, hash) from main() and fetch the full data via artifacts.

Example fix

# before (executed code)
print(json.dumps(huge_payload))

# after (executed code)
from pathlib import Path
Path('artifacts/output.json').write_text(json.dumps(huge_payload))
print(json.dumps({'status': 'ok', 'rows': len(huge_payload)}))

Defensive patterns

Strategy: try-catch

Try / catch

try:
    result = provider.execute_code(instance_id, code, "python")
except RuntimeError as e:
    if "output exceeded" in str(e):
        # rerun with quieter code or a higher cap; treat as workload error, not infra error
        raise OutputTooLarge(str(e)) from e
    raise

Prevention

Instruct generated code to avoid printing raw large payloads; use artifacts files instead.
Silence chatty stderr sources (progress bars, pip, warnings) in the child environment.
Set max_output_bytes proportional to expected output and monitor near-limit runs.

When it happens

Trigger: Generated code that prints large dataframes, long lists, base64 blobs, or verbose install/logging output; combining a chatty library (pip, transformers, debug logging) on stderr with prints on stdout; a low max_output_bytes configured at initialize().

Common situations: Agent-written code doing print(df) or print(json.dumps(big_payload)); ML libraries emitting progress bars/warnings to stderr; forgetting that stderr counts toward the same budget; setting max_output_bytes near the 1024 minimum.

Related errors

AI-assisted analysis of infiniflow/ragflow@554fb1133a (2026-08-15). Data as JSON: /api/errors/f531ef78ebd8e797. Report an issue: GitHub.