ec2-purge-snapshots

What is it?

A Ruby script that lets you purge (delete) AWS EC2 snapshots according to rules you set up.

For example,

./ec2-purge-snapshots.rb -v all -h 48 -d 14 -w 4 -m 24

means "keep every snapshot on all volumes for 48 hours, one per day for two weeks, one per week for four weeks, one per month for two years - delete everything else".

The idea is that you have another job which takes a snapshot, say, every hour, then run this script periodically to clean up.

Usage

Usage: ./ec2-purge-snapshots.rb [options]

Deletes ALL snapshots (for the volumes specified) that do not
match the rules below. Rules are applied in the following order:

    hours -> days -> weeks -> months

MANDATORY options (one of -v or -t must be used):
    -v, --volumes VOL1,VOL2,...      Comma-separated list (no spaces) of volume-ids,
                                     or 'all' for all volumes
    -t, --tag KEY=VALUE              Tag to use to filter the snapshot. May specify multiple tags.

MANDATORY rules:
    -h, --hours HOURS                The number of hours to keep ALL snapshots
    -d, --days DAYS                  The number of days to keep ONE snapshot per day
    -w, --weeks WEEKS                The number of weeks to keep ONE snapshot per week
    -m, --months MONTHS              The number of months to keep ONE snapshot per month

OPTIONAL options:
    -a, --access-key-file FILENAME   The path to a file containing the AWS access key to use,
                                     otherwise use the value of $AWS_ACCESS_KEY
    -e, --secret-key-file FILENAME   The path to a file containing the AWS secret key to use,
                                     otherwise use the value of $AWS_SECRET_KEY
    -n, --noop                       Don't actually delete, but print what would be done
    -q, --quiet                      Print deletions only
    -s, --silent                     Print summary only
        --no-summary                 Don't print summary
    -x, --extremely-silent           Don't print anything unless something goes wrong
    -u, --url URL                    The Amazon EC2 region URL (default is US East 1)
        --help                       Show this message

Known issues

• If you have many snapshots the script will eventually return the AWS error RequestLimitExceeded. Not sure how to fix this, since there is no information given about which limit is exceeded - maybe just add a slight delay between each API call?

Contribute

Please fork and add pull requests if you would like to improve this package.

Author

Stian Grytøyr