Python 3 Script to Upload Parallelly Multiple Files to Amazon S3 Bucket Cloud Storage in Command Line



s3-parallel-put Parallel uploads to Amazon AWS S3

s3-parallel-put speeds the uploading of many small keys to Amazon AWS S3 by executing multiple PUTs in parallel.




The program reads your credentials from the environment variables AWS_ACCESS_KEY_ID and AWS_SECRET_ACCESS_KEY.

Keys are computed by combining PREFIX with the path of the file, starting from SOURCE. Values are file contents.


  • -h, --help — show help message
S3 options:
  • --bucket=BUCKET — set bucket
  • --bucket_region=BUCKET_REGION — set bucket region if not in us-east-1 (default new bucket region)
  • --host=HOST — set AWS host name
  • --secure and --insecure control whether a secure connection is used
Source options:
  • --walk=MODE — set walk mode (filesystem or tar)
  • --exclude=PATTERN — exclude files matching PATTERN
  • --include=PATTERN — don’t exclude files matching PATTERN
  • --ignore-files-older-than-days=DAYS — ignore files older than defined DAYS, only sync newer files
Put options:
  • --content-type=CONTENT-TYPE — sets the Content-Type header, set to “guess” to guess based on file name or “magic” to guess by filename and libmagic

  • --gzip — compresses common text files and sets the Content-Encoding header to gzip

  • --gzip-type=GZIP_TYPE — if --gzip is set, sets what content-type to gzip, defaults to a list of known text content types, “all” will gzip everything. Specify multiple times for multiple content types (eg.--gzip-type=guess --gzip-type="image/svg+xml") [default: “guess”]

  • --put=MODE — sets the heuristic used for deciding whether to upload a file or not. Valid modes are:

    • add set the key’s content if the key is not already present.
    • stupid always set the key’s content.
    • update set the key’s content if the key is not already present and its content has changed (as determined by its ETag).
    • copy
      The default heuristic is update. If you know that the keys are not already present then stupid is fastest (it avoids an extra HEAD request for each key). If you know that some keys are already present and that they have the correct values, then add is faster than update (it avoids calculating the MD5 sum of the content on the client side).
  • --prefix=PREFIX — set key prefix

  • --resume=FILENAME — resume from log file

  • --grant=GRANT — A Canned ACL policy to be applied to each file uploaded. Choices: privatepublic-readpublic-read-writeauthenticated-readbucket-owner-readbucket-owner-full-controllog-delivery-write

  • --header=HEADER:VALUE — adds an arbitrary header to the S3 file. This option can be specified multiple times.

  • --encrypt-key — use server side encryption

Logging options:
  • --log-filename=FILENAME — set log filename
  • -q, --quiet — less output
  • -v, --verbose — more output to be printed, including progress of individual files
Debug and performance tuning options:
  • --dry-run — don’t write to S3. Causes the program to print what it would do, but not to upload any files. It is strongly recommended that you test the program with this option before transferring any real data.
  • --limit=N — set maximum number of keys to put. Causes the program to upload no more than N files. Combined with --dry-run, this is also useful for testing.
  • --processes=N — sets the number of parallel upload processes



Leave a Reply