Skip to navigation

aistore.sdk.blob_download_config

View as MarkdownOpen in Claude

Module Contents

Classes

NameDescription
BlobDownloadConfigConfiguration for downloading objects using a blob downloader

API

class aistore.sdk.blob_download_config.BlobDownloadConfig(
chunk_size: str = None,
num_workers: str = None,
read_timeout: str = None,
threshold: int = None
)
Dataclass

Configuration for downloading objects using a blob downloader

Example

blob_settings = BlobDownloadConfig( threshold=64 * 1024 * 1024, # Download objects of at least 64 MiB chunk_size=“1MiB”, # 1 MiB per chunk num_workers=“5”, # 5 concurrent download workers read_timeout=”30s” # Per-attempt backend read timeout )

Objects smaller than 128 KiB, or fitting in a single chunk, are served via regular GET (no blob-download job).

chunk_size
str = None

Chunk size for the blob downloader. It can be specified in IEC or SI units, or as raw bytes (e.g., “4mb”, “1MiB”, “1048576”, “128k”)

num_workers
str = None

Number of concurrent workers for the blob downloader: zero (default) or at least two

read_timeout
str = None

Per-attempt timeout for reading a chunk from the backend, expressed as a duration string (e.g., ”30s”, “2m”)

threshold
int = None

Minimum remote object size in bytes that enables the blob downloader. Zero disables blob download; omitting it preserves forced blob download.

aistore.sdk.blob_download_config.BlobDownloadConfig.__post_init__()
aistore.sdk.blob_download_config.BlobDownloadConfig.as_headers() -> dict

Return request headers for this blob-download configuration.