abhigyanpatwari/GitNexus · error · OSError

short write while staging oracle

Error message

short write while staging oracle

What it means

The only OSError in this set, raised inside _write_stage_file when os.write() returns <= 0 during the staged write loop. Writes to a regular file should never return 0; a non-positive return indicates a filesystem-level failure (ENOSPC, EIO, quota, or a full tmpfs/NFS write boundary).

Source

Thrown at eval/workflow_bench/oracle_assets.py:482

    destination.parent.mkdir(parents=True, mode=0o700, exist_ok=True)
    current = stage_root
    for part in PurePosixPath(item.target).parts[:-1]:
        current /= part
        metadata = current.lstat()
        if stat.S_ISLNK(metadata.st_mode) or not stat.S_ISDIR(metadata.st_mode):
            raise ValueError(f"oracle stage parent must be a real directory: {item.target}")
        current.chmod(0o700)
    descriptor = os.open(
        destination,
        os.O_WRONLY | os.O_CREAT | os.O_EXCL | getattr(os, "O_NOFOLLOW", 0),
        0o400,
    )
    try:
        view = memoryview(item.payload)
        while view:
            written = os.write(descriptor, view)
            if written <= 0:
                raise OSError("short write while staging oracle")
            view = view[written:]
        os.fchmod(descriptor, 0o400)
        os.fsync(descriptor)
    finally:
        os.close(descriptor)


def _verify_staged_oracle(stage_root: Path, snapshot: TaskOracleSnapshot) -> None:
    root_metadata = stage_root.lstat()
    if stat.S_ISLNK(root_metadata.st_mode) or not stat.S_ISDIR(root_metadata.st_mode):
        raise ValueError("oracle stage root changed during verification")
    for item in snapshot.files:
        relative = PurePosixPath(item.target)
        current = stage_root
        for part in relative.parts[:-1]:
            current /= part
            metadata = current.lstat()
            if stat.S_ISLNK(metadata.st_mode) or not stat.S_ISDIR(metadata.st_mode):

View on GitHub (pinned to d540b00184)

Solutions

  1. Check free space and quota: `df -h <worktree>` and `quota -s` (if applicable).
  2. Point the worktree at a local filesystem with headroom >= MAX_ORACLE_TOTAL_BYTES.
  3. Ensure the stage root is a real directory and the file is a regular file (see [358]) before writing.
  4. Retry staging on a clean FS after freeing space; if it recurs, treat the FS as unhealthy.
Defensive patterns

Strategy: try-catch

Validate before calling

import os, shutil
from pathlib import Path
from eval.workflow_bench.oracle_assets import MAX_ORACLE_TOTAL_BYTES

def stage_has_headroom(worktree: Path) -> bool:
    usage = shutil.disk_usage(worktree)
    return usage.free >= MAX_ORACLE_TOTAL_BYTES

Type guard

def is_oracle_short_write(exc: BaseException) -> bool:
    return isinstance(exc, OSError) and "short write" in str(exc)

Try / catch

try:
    with oracle_assets.staged_task_oracle(worktree, snapshot) as root:
        run_command(root)
except OSError as exc:
    raise AbortTask(f"oracle staging I/O failed: {exc}") from exc

Prevention

When it happens

Trigger: Triggered when os.write(descriptor, view) returns 0 or negative while staging oracle bytes into .wfbench-oracle-<hex>/<target> — disk full, quota exceeded, NFS/overlayfs write error, or the staging FS was a pipe/special file substituted via a race.

Common situations: Out-of-disk on the worktree volume; a small tmpfs used for staging; a quota limit hit mid-write; overlayfs ENOSPC; a model process racing to replace the file with a pipe.

Related errors

AI-assisted analysis of abhigyanpatwari/GitNexus@d540b00184 (2026-08-12). Data as JSON: /api/errors/f4653c41c9e8a64d. Report an issue: GitHub.