Handling automatic restarts of unicorn in an HA environment
When upgrading from one Ruby version to another, it seems that we automatically restart unicorn (https://gitlab.com/gitlab-org/omnibus-gitlab/blob/26abf736c1a5899462bb5ffc552cb463ee5e1e3c/files/gitlab-cookbooks/gitlab/recipes/gitlab-rails.rb#L382-385).
This could lead to 502s on the server until unicorn comes back up.
In general, the restart makes sense as long as you're not running in an HA environment or don't require no-downtime upgrades.
I'm raising this issue because if we upgrade to 2.4.4 in #2588 (closed), then we will automatically restart. I'm surprised we didn't run into this when we upgraded from 2.3.6 to 2.3.7.
Do we need some other option that disables the automatic restart of unicorn and other services?
Edited by Stan Hu