jspiewak / s3_disk_util

Like `du` but for S3

s3_disk_util

• What -- A tool that allows a user to visualize which buckets (and parts of buckets) are using the most data storage.
• Why -- Because I'm trying to pare down my S3 bill, and the S3 control panels (even CloudWatch) do not really provide anything similar.
• Inspiration -- This script is meant to be like the du tool for linux, except for inspecting the disk usage of s3 buckets.
• How -- It will traverse s3 buckets and provide high level disk usage information to stdout.

Usage

kevinowocki@local /Users/kevinowocki/Desktop/s3_disk_util~ % python3 du.py --help
usage: du.py [-h] [-b BUCKET] [-p PROFILE] [-d DEPTH] [-di DIR]

This script is meant to be like the `du` tool for linux, except for inspecting
the disk usage of s3 buckets. It will traverse s3 buckets and provide high
level disk usage information to stdout.

optional arguments:
  -h, --help            show this help message and exit
  -b BUCKET, --bucket BUCKET
                        Bucket to examine (ex: 'com.owocki.assets')
  -p PROFILE, --profile PROFILE
                        AWS credentials profile name (default: 'default')
  -d DEPTH, --depth DEPTH
                        Depth to examine bucket (ex: 4)
  -di DIR, --dir DIR    Directory to examine (ex: 'logs/')

Example

kevinowocki@local /Users/kevinowocki/Desktop/s3_disk_utils~ % python3 du.py --depth=1 --bucket=BUCKETNAME --profile=mytestaccount
BUCKETNAME
(Cloudwatch bucket size estimate: 22.7GiB)
/ : 22.7GiB
- DIR1/ : 22.6GiB
- DIR2/ : 452.6KiB
- DIR3/ : 1.6MiB
- DIR4/ : 119.0MiB
- DIR5/ : 0.0B

kevinowocki@local /Users/kevinowocki/Desktop/s3_disk_util~ % python3 du.py --depth=2 --bucket=BUCKETNAME
BUCKETNAME
(Cloudwatch bucket size estimate: 22.7GiB)
/ : 22.7GiB
- DIR1/ : 22.6GiB
-- DIR1/SUBDIR1/ : 31.1MiB
-- DIR1/SUBDIR2/ : 12.7GiB
-- DIR1/SUBDIR3/ : 0.0B
-- DIR1/SUBDIR4/ : 9.9GiB
- DIR2/ : 452.6KiB
-- DIR2/SUBDIR1/ : 429.5KiB
- DIR3/ : 1.6MiB
-- DIR3/SUBDIR1/ : 254.4KiB
- DIR4/ : 119.0MiB
- DIR5/ : 0.0B

Setup

1. Create a AWS IAM user at https://console.aws.amazon.com/iam/home.
   • Make sure your user has AmazonS3FullAccess and CloudWatchReadOnlyAccess policies.
2. Use your existing ~/.aws/credentials file and profile names or create a config file that looks like this:

kevinowocki@local /Users/kevinowocki/Desktop/s3_disk_utils~ % cat ~/.aws/credentials
[default]
aws_access_key_id = ACCESS_KEY_GOES_HERE
aws_secret_access_key = SECRET_KEY_GOES_HERE
region=REGION

3. Clone this repo.
4. Install python3 (if needed) and boto3 (if needed).
   • To instally python3, instructions different depending upon your OS. Here are some instructions for Mac OS X
   • To install boto3, pip install -r requirements.txt
5. Run du.py with the usage described above.

What else

This script can run a little slow on larger buckets. Thats okay; This is a limitation inherent to the way this information is provided via AWS APIs. Pipe du.py's' output to a file (perhaps inside of a screen or tmux) and come back later.