This project is deprecated, and should not be used. 2023-08-01
tummy-backup is as simple a possible, but no simpler. The core of it is based around simple, reliable routines: ZFS, rsync, cron+xargs for the scheduler and parallel execution. And at it's heart is a shell script wrapping rsync that I've been refining since the '90s.
Following the Unix philosophy, it is quiet to indicate success, and makes noise when there are problems. It does include a fully-featured web management interface, including backup browser and file/directory recovery.
ZFS does all the heavy lifting of managing historical data. Unlike other rsync-based backups, you won't fill the system with inodes or have duplicates of large files that are appended to (log files, Zope databases, any journal and most database files).
This is a backup system that was developed by tummy.com and has been in use since 2010. We have deployed it among a number of our consulting clients, handling backups of between 5 and 200 servers. It provides a central management interface, so managing 1 backup server is as easy as managing 10.
tummy-backup has proven to be an extremely robust and reliable system over the years.
It should work on any platform that has: rsync, ZFS, Python, and Postgres.
I know some have deployed it to CentOS. I don't currently have any CentOS systems, so the only instructions I have for installing are for Ubuntu-like OSes. See the "INSTALL" files in this directory for Ubuntu 12.04 (also 16.04, possibly 18.04), and Ubuntu 20.04 instructions.
- Web management interface.
- Space-efficient storage of deltas, deduplication and compression via ZFS.
- Secure backups using single-purpose SSH keys.
- Low maintenance: e-mail is only sent when attention is needed.
- Robust and reliable: Thousands of servers have been backed up for more than a decade.
- Clustered: Managing multiple backup servers is as easy as managing one.