I’m running GitLab on EC2 t2.micro instance. This installation is solely used only for backup purpose.
I’m aware of the GitLab S3 backup configuration, which works fine, but does a full backup (not differential). I’m using this for daily backup.
To have a more fine grained differential backup outside my premise, I’ve taken a EC2 t2.micro machine with GitLab and using cron job I keep doing
git pull origin and
git push backupserver --mirror every 5 minutes.
I also understand that my machine’s hardware resources are less than the recommended requirements for GitLab CE. This is where I need help.
After doing the following change I’m able access GitLab portal more reliably:
unicorn[‘worker_timeout’] = 120
unicorn[‘worker_processes’] = 1
Before this change, I usually got 502 error in my portal.
Is there some other configuration that I can change to make this experience more smooth?
For example, GitLab runs 7 processes: gitlab-workhorse, logrotate, nginx, postgresql, redis, sidekiq, unicorn. Since I’m using only the “git repository” feature on this machine, may be I can stop some of the services that are not required for my backup machine.
Currently, sometimes while doing
git push backupserver I get timeout message. I’m not worried if git push is slow, but it should not fail.