GitLab Omnibus: downtime on a single node installation?

Hi all,

the GitLab Docs state:

“For single node installations, GitLab is not available to users while an upgrade is in progress. The user’s web browser shows a Deploy in progress message or a 502 error”

Ok, when i do my upgrades i always start with

gitlab-ctl deploy-page up

But just because i was curios i tried to push a local commit to one of my repos on the server during upgrade, and this worked. This isn’t really what i expected, and i wonder if this is safe or if i have to take more actions to prevent my users from doing this when i upgrade?

How do you push? Over http/https or SSH? If over SSH, then the SSH daemon is ran via the operating system and not Gitlab. Since an ssh key will be uploaded for this, this is the most likely reason why since it won’t need to check/verify via Gitlab.

On my installation, we use just HTTP/HTTPS for push/pull, and when I do my upgrades nobody can push since the Gitlab services aren’t running to allow it.

I pushed via HTTPS, and it worked.

Had gitlab stopped all the services? For example on Debian/Ubuntu, when the package installation has started, it takes a while before it gets to the bit where it stops all Gitlab services (other than postgres so that it can do db migrations). So technically you could have still been able to push because of this.

I don’t know if all services where stopped. Package installation had started, i assume, since the progress bar was shown.

Ok, so i assume that GitLab stops its services ‘gracefully’ during upgrades, allowing pending transactions to complete. Thanks :slight_smile: