I'm using Gitlabs latest Omnibus-package on an EC2 Ubuntu machine.
To refresh my SSL certificate (issued via Let's Encrypt) I need to stop Gitlab's Nginx so Let's Encrypt can verify that I possess the domain.
Therefore I hit sudo gitlab-ctl stop.
The sudo gitlab-ctl status afterwards is:
down: gitlab-workhorse: 325s, normally up; run: log: (pid 1109) 5361843s
down: logrotate: 324s, normally up; run: log: (pid 1104) 5361843s
down: nginx: 324s, normally up; run: log: (pid 1103) 5361843s
down: postgresql: 324s, normally up; run: log: (pid 1101) 5361843s
down: redis: 323s, normally up; run: log: (pid 1102) 5361843s
down: sidekiq: 322s, normally up; run: log: (pid 1112) 5361842s
down: unicorn: 322s, normally up; run: log: (pid 1100) 5361843s
However when I access my domain I get Nginx' 502 Bad Gateway.
How can I truly stop its internal Nginx.
Besides the certificate part the etc/nginx/gitlab.rb is still the default.
Here's the output of ps -eaf|grep -i nginx
root 1091 985 0 2015 ? 00:07:15 runsv nginx
root 1103 1091 0 2015 ? 00:04:14 svlogd -tt /var/log/gitlab/nginx
gitlab-+ 24669 1 0 2015 ? 01:03:38 nginx: worker process
root 27272 1091 0 13:12 ? 00:00:00 /opt/gitlab/embedded/sbin/nginx -p /var/opt/gitlab/nginx
ubuntu 27275 27254 0 13:12 pts/2 00:00:00 grep --color=auto -i nginx
For completeness sake what I ended up doing three years ago was what @user8215365 suggested.
Simply invoking sudo gitlab-ctl stop nginx did the trick.