Handle outdated s3 backups

[aitor]

Lately I have been using s3 backups as described in the Roots guide:

Backing Up Trellis Sites to an S3 Bucket | Roots

It works fine, but I don’t know how to remove old backups from s3 bucket. I have no experience with s3 and I tried to manage it with a lifecycle setup, unsuccesfully. Is it posible to remove automatically old backups? (for example, keep the latest 3 backups).

[austin]

You could use lifecycle methods to get rid of old stuff or transition old stuff to cheaper storage tiers or just delete them

https://docs.aws.amazon.com/AmazonS3/latest/dev/object-lifecycle-mgmt.html

[aitor]

Thank you for the answer! Ok, it is possible, but i prefer to keep n latest files instead expires based on time. The problem I can see with expiration after n days is, if the server stops making backups (for any reason), all backups will end up expiring and disappearing.

I’m reading docs. I will come back to solve thread if I find the right way to solve it.

[aitor]

The simplest way I found to keep at least one copy of each backup is:

1. Remove timestamp from file name:

Replace

# site/scripts/backup-to-s3.sh
ARCHIVE_FILENAME=$SITE-$ENVIRONMENT-$TIMESTAMP.tar.gz

by

# site/scripts/backup-to-s3.sh
ARCHIVE_FILENAME=$SITE-$ENVIRONMENT.tar.gz

2. Enable version control for the bucket.

3. Set bucket lifecycle with expiration time for previous versions and leave untouched current version.

[austin]

Yep looks good to me!