Note that on a single-core server it may take up to a minute to restart Unicorn and Sidekiq. Your GitLab instance will give a 502 error until Unicorn is up again.
It is also possible to start, stop or restart individual components.
sudo gitlab-ctl restart sidekiq
Unicorn supports zero-downtime reloads. These can be triggered as follows:
sudo gitlab-ctl hup unicorn
Note that you cannot use a Unicorn reload to update the Ruby runtime.

Try using the above two commands and find that it doesn't work. When I got angry, I kept typing gitlab-ctl status until I found that unicorn pid was getting bigger all the time. The PID for the other services has not changed.
3. Now we've almost found the problem -- unicorn's question. For the official tutorial, you can use gitlab-ctl tail unicorn to track the unicorn's status. Unfortunately, when we discovered that port 8080 was occupied, the unicorn's status was reported to have been captured

E, [2015-02-11T17:27:57.818492 #26687] ERROR -- : adding listener failed addr=127.0.0.1:8080 (in use)
E, [2015-02-11T17:27:57.818621 #26687] ERROR -- : retrying in 0.5 seconds (4 tries left)
E, [2015-02-11T17:27:58.318902 #26687] ERROR -- : adding listener failed addr=127.0.0.1:8080 (in use)
E, [2015-02-11T17:27:58.318998 #26687] ERROR -- : retrying in 0.5 seconds (3 tries left)
E, [2015-02-11T17:27:58.819309 #26687] ERROR -- : adding listener failed addr=127.0.0.1:8080 (in use)
E, [2015-02-11T17:27:58.819423 #26687] ERROR -- : retrying in 0.5 seconds (2 tries left)
E, [2015-02-11T17:27:59.319954 #26687] ERROR -- : adding listener failed addr=127.0.0.1:8080 (in use)
E, [2015-02-11T17:27:59.320076 #26687] ERROR -- : retrying in 0.5 seconds (1 tries left)

4. Well, here's the problem. The choice then became whether to kill the original 8080 port service or to change the unicorn port. This depends on your specific needs. I've changed the unicorn port to 9090, and the method is the one described at the beginning.