michael schurter

Rsync is an excellent file transfer utility thats especially well suited for backing up files over the Internet because it only transfers the data that has changed. A friend asked me how to set it up, so I thought I’d post what I sent him here.

Goal: Backup a directory from computer Zim to computer Ark

Details:

• Both Zim and Ark are subdomains of example.com
• The user on Ark which receives the backup files is named backupuser
• The user on Zim with access to the files you want to backup is named steve

Prerequisites:

• ssh installed on both hosts
• rsync installed on both hosts

1. Login to Zim via ssh:

   ssh steve@zim.example.com

2. Generate a ssh key pair using:

   ssh-keygen -t rsa
   <press enter when prompted where to save the key>
   <press enter twice when asked for a passphrase>

3. To use the key to login to Ark remotely without manually entering a password you need to copy the public key from Zim to Ark using:

   ssh-copy-id -i .ssh/id_rsa.pub backupuser@ark.example.com

   If you don’t have ssh-copy-id on your system, get a new system. If thats not possible you can download the script with:

   wget -O ssh-copy-id http://cvsweb.mindrot.org/index.cgi/~checkout~/openssh/contrib/ssh-copy-id?rev=1.6;content-type=text%2Fplain && chmod +x ssh-copy-id

   Then retry the above command only you’ll need to prepend a “./”:

   ./ssh-copy-id -i .ssh/id_rsa.pub backupuser@ark.example.com

4. Verify the key copied properly by attempting to login to Ark. You should not be prompted for a password:

   ssh backupuser@ark.example.com

5. Logout of Ark. The key is setup, so you’re now ready to rsync files without having to manually enter a password.
6. Test rsync by choosing a small file to backup and using:

   rsync -tP /some/small/testfile backupuser@ark.example.com:/tmp

   A nice little progress bar should be displayed as the file is transferred. Confirm that “testfile” is now in /tmp on Ark.

7. You’re finally ready to do a real rsync like:

   rsync -t /directory/to/backup/* backupuser@ark.example.com:/existing/backup/directory

   Note: There are several useful options for rsync. Check man rsync to find out more.

   • -p — preserve permissions (useful for backups, use -E if you only care about the executable bit)
   • -r — recursively backup directories.
   • -z — compressed uncompressed files
   • And just FYI: -t tells rsync to use the last modified timestamp to determine whether or not to transfer files. It makes rsync a lot faster at determining whether or not files have changed.
8. To schedule the backup to take place nightly at 1:13 AM edit your crontab using crontab -e and insert the following line:

   13 1 * * * rsync -qt /directory/to/backup/* backupuser@ark.example.com:/existing/backup/directory

Caveats:

• These instructions will push files from Zim to Ark. There’s no reason why Ark couldn’t pull files from Zim. In fact, this is often more secure if Zim is a web server with a larger attack surface than Ark. Mea culpa.
• If the IP address of Ark is dynamic, use a service like dyndns.com. Otherwise SSH will give you errors.
• Major security warning: If someone breaks into Zim, they can also delete all of your backups on Ark. Never ever ever use the root user for backups on Ark. You can use the root user on Zim to send the backups, but its best to have a special backup user setup on Ark to receive the backup.