Automating the Backup of GitHub Repositories Using ghbackup

I explain my setup to keep backups of my repositories on my own server.

As a software developer GitHub has become a central part of my work. The GitHub crew does a great job. But even for the best companies it's possible to get in trouble some day.

Instead of setting all on a single company, it's a good idea to take advantage of the distributed nature of Git and keep your own backups for emergency situations.
This can be easily automated so you don't have to worry about it anymore.

Here is how I did it:

I decided to setup a small DigitalOcean server to backup all my repositories. The latest version is fetched every night and I get notified in Slack. When the notifications ever stop coming, I know something is wrong with my backups.

I use ghbackup to do this. With ghbackup I can backup all public and private repositories I have access to in a single go, and if a project owner would decide to remove me from a project I would still have a copy of the repositories.

Follow these steps to create a similar setup or adopt it to your own needs:

We'll be using ghbackup, sleepto and systemd.

• First, you need a GitHub token with at least the repo scope and you need a Slack Webhook URL.

• All following steps happen on system level. To not prefix every command with sudo, let's login as root:
  

sudo -s

• systemd is already running on most Linux systems. Install ghbackup and sleepto. In this example we keep executables in the directory /opt

curl -sL https://github.com/qvl/sleepto/releases/download/v1.6/sleepto_1.6_linux_64bit.tar.gz | tar -xzf - -C /opt sleepto
curl -sL https://github.com/qvl/ghbackup/releases/download/v1.8/ghbackup_1.8_linux_64bit.tar.gz | tar -xzf - -C /opt ghbackup

• Create a wrapper script to run the backup. Update the variables below with your own values:

touch /opt/ghbackup-job && chmod +x $_ && echo > $_ '#!/usr/bin/env bash

github_token=""
slack_hook=""
backup_dir="/srv/github-backup"

/opt/ghbackup -secret $github_token $backup_dir \
  && curl -X POST --data-urlencode \
  "payload={\"text\": \"Successful GitHub backup.\"}" \
  $slack_hook
'

• Create a new unit file to run the created ghbackup-job as a repeating service. We use sleepto to trigger it every day at 1am.

touch /etc/systemd/system/ghbackup.service && chmod 644 $_ && echo > $_ '
[Unit]
Description=Github backup
After=network.target

[Service]
ExecStart=/opt/sleepto -hour 1 /opt/ghbackup-job
Restart=always

[Install]
WantedBy=multi-user.target
'

• Now the only thing left to do is to enable this service. We make sure it's automatically started when the system boots:

systemctl daemon-reload
systemctl start ghbackup
systemctl enable ghbackup

• The service should be running now:

systemctl status ghbackup

If everything works you should get notifications about successful backups soon!

You can also trigger sleepto to run the backup job immediately:

systemctl kill -s ALRM ghbackup

After the first backup completed you find the repositories in /srv/github-backup.

This is how I setup my GitHub backups. I hope this is helpful to others as well.

Please let me know if you run into any problems!

Note: This post was originally posted on jorin.me.

Comments

[slashtechno]

This tool seems amazing! I made a simple Github backup script as well, but repo names needed to be entered manually. Glad there is a program to replace my current solution.