unicorn.rb.example.development. Also remove unneeded configurations
and update the comment accordingly. I think REE was long dead.

The Unicorn master resets any signal handlers and prevents rbtrace
from working properly as a result. Move the `require 'rbtrace'` call
to the Unicorn `after_fork` block and keep the initializer only for Sidekiq.

Issuing a SIGUSR2 to a unicorn involves a dance where new workers are
spun up before the old ones are shut down. To ensure Prometheus
.db files aren't deleted after the new workers are shut down, we
need to clean these files up before the unicorn process is run.

Relates to https://gitlab.com/gitlab-org/omnibus-gitlab/merge_requests/2582

The service discovery mechanism can be used to automatically update the
list of database secondaries. This is done by periodically checking a
DNS record, updating the load balancer whenever its list of hosts
differs from the DNS record.

Old connections are terminated after the new hosts have been added to
the load balancer. A timeout mechanism is put in place to ensure this
completes in a fixed amount of time. Given requests check their
connections back in at the end of a request, the usual time this takes
should be in the order of a few seconds.

unicorn forking new workers.

[ci skip]

[ci skip]

This reverts commit 516bcabb.

Conflicts:
	Gemfile

A value of 1 cause http push fail (issue #6978 on GitHub)

Signed-off-by: Dmitriy Zaporozhets <dmitriy.zaporozhets@gmail.com>

- If you don't edit gitlab-shell/config.yml, you wnon't be able to push via SSH (as described here: http://stackoverflow.com/questions/15746336/gitlab-cant-push-code-to-server)
- If you don't re-run sudo -u git -H bundle exec rake assets:precompile RAILS_ENV=production, some icons won't be visible

Added reference to the line "ENV['RAILS_RELATIVE_URL_ROOT'] = "/gitlab"" , as it talked about in the gitlab.yml.example

This commit addresses two related issues:
- Sidekiq should shut down completely before we restart it;
- If we restart unicorn with USR2 we need to kill the old master and
  worker processes.

otherwise it will listen on 0.0.0.0

Most of these are comments but a few are strings for users.

Might be an idea to run this from time to time:
https://github.com/lyda/misspell-check

It runs mostly clean now.

The timeout was changed to 30 and the comment wasn't updated. Also I've seen several reports of this timeout not being long enough for quite a few people. It resulted in me getting a 502 bad gateway from nginx until I upped it. Perhaps could back to the default 60 would be better? I changed mine to 300, running on an AWS micro instance.

Renamed unicorn.rb.orig to unicorn.rb.example since this seems to be more in line with the common used naming in the config directory. Plus seeing *.orig files I think of the leftovers of git merges, which left me puzzled for quite some time.