Gunicorn
Following up on what has been done for Octobus mercurial mirror, we should use a production grade WSGI server, such as Gunicorn, instead of the current hg serve
.
This is for internal Mercurial serving currently exposed to gitlab-rails
. In some farther future, it could be exposed to gitlab-workhorse
or to a Gitaly protocol implementation when the time comes for it.