For myself, running the latest Gitlab CE container, periodically ever 3-4 hours I encounter the gitlab-rails/production logged error:
Completed 500 Internal Server Error in 15670ms (ActiveRecord: 152.5ms)
Errno::EAGAIN (Resource temporarily unavailable - connect(2) for /var/opt/gitlab/redis/redis.socket): config/initializers/rack_attack_git_basic_auth.rb:7:in `block in <top (required)>' lib/gitlab/request_context.rb:18:in `call'
Looking at all existing topics for issue with this redis socket, either doing a gitlab-ctl reconfigure (which I’ve tried) or a stop/start of redis resolved their issues but for me it continues to periodically occur in the timeframe of every 3-4 hours.
I see no errors with the Redis container, only the gitlab container and I haven’t yet seen any additional logging that can tell me exactly why there’s a disconnect.
I saw from this Gitlab topic: https://gitlab.com/gitlab-org/gitlab-ce/issues/14428
that his solution to this exact error was running some manual command but trying didn’t work for me either and the ticket was closed. Would this perhaps be an existing bug with the latest DEE (Docker Enterprise Edition) and GitlabCE? Thank you for your time.