unslothai/unsloth · info · S3DownloadCancelled

S3 dataset download cancelled

Error message

S3 dataset download cancelled

What it means

S3DownloadCancelled (a RuntimeError subclass, s3_dataset.py:39) raised by _raise_if_cancelled when the caller-supplied cancel_callback returns True. The download loop polls the callback at each phase (client build, listing, per-object download) so long S3 transfers can be aborted cooperatively. It is a control-flow signal, not a failure — the caller requested the cancellation.

Source

Thrown at studio/backend/core/training/s3_dataset.py:145

        f"({', '.join(families)}). Keep one dataset format under the selected prefix."
    )


def _unique_local_path(target_dir: str, filename: str, used_paths: set[str]) -> str:
    """Return an unused flattened path for an S3 object basename."""
    stem, ext = os.path.splitext(filename)
    candidate = os.path.join(target_dir, filename)
    suffix = 1
    while candidate in used_paths or os.path.exists(candidate):
        candidate = os.path.join(target_dir, f"{stem}_{suffix}{ext}")
        suffix += 1
    used_paths.add(candidate)
    return candidate


def _raise_if_cancelled(cancel_callback: Optional[Callable[[], bool]]) -> None:
    if cancel_callback is not None and cancel_callback():
        raise S3DownloadCancelled("S3 dataset download cancelled")


def prepare_s3_dataset_download(
    s3_config: dict,
    dest_dir: Optional[str] = None,
    cancel_callback: Optional[Callable[[], bool]] = None,
) -> S3DatasetDownload:
    """Download supported dataset files from S3 to a local directory.

    Returns the local files plus the owned temporary directory, when one was
    created. Call ``cleanup()`` after the dataset loader has materialized data.

    Raises ``RuntimeError`` if boto3 is missing, and ``ValueError`` if the
    bucket/prefix contains no supported dataset files.
    """
    if not boto3_available():
        raise RuntimeError("S3 dataset loading requires boto3. Install it with: pip install boto3")

View on GitHub (pinned to 203007d190)

Solutions

1. Catch S3DownloadCancelled explicitly in the caller and treat it as a normal cancellation (stop the workflow, clean the temp dir via the returned object's cleanup() or your own dest_dir).
2. If the cancellation was unintended, fix the callback's state — e.g. a shared flag that a previous failed run left set to True.
3. Reset/clear the cancel flag before starting a new download.

Example fix

// before
try:
    dl = prepare_s3_dataset_download(cfg, cancel_callback=cb)
except RuntimeError as e:  # swallows cancellation as generic failure
    log.error(e)
// after
from core.training.s3_dataset import S3DownloadCancelled
try:
    dl = prepare_s3_dataset_download(cfg, cancel_callback=cb)
except S3DownloadCancelled:
    log.info("Download cancelled by user")
    return

if cancel_callback is not None and cancel_callback():
    # skip the call entirely — user already cancelled
    return None

from core.training.s3_dataset import S3DownloadCancelled

try:
    dl = prepare_s3_dataset_download(cfg, cancel_callback=cb)
except S3DownloadCancelled:
    # expected control flow: abort the training job cleanly
    abort_job(reason='download_cancelled')
    return

When it happens

Trigger: Passing cancel_callback=lambda: True (or one that flips to True after the user clicks Stop) into prepare_s3_dataset_download; the exception surfaces at the next callback checkpoint — during listing or between object downloads.

Common situations: UI 'Cancel' button wired to a flag the download polls; job orchestration cancelling a queued training run; timeouts implemented by the caller via a deadline-based callback.

Related errors

• Diffusion generation was cancelled.
• Audio generation cancelled
• Transcription cancelled.
• S3 prefix contains mixed dataset formats ({', '.join(familie
• S3 dataset loading requires boto3. Install it with: pip inst

AI-assisted analysis of unslothai/unsloth@203007d190 (2026-08-15). Data as JSON: /api/errors/511e4c28fd7881a0. Report an issue: GitHub.