Use Omnibus new syntax for building projects.

See merge request !404

Gitlab shell secret symlinking

See merge request !400

Gitlab shell secret symlinking

See merge request !400

This reverts commit 1911109c, reversing
changes made to a8cdae1b.

This reverts commit 1911109c, reversing
changes made to a8cdae1b.

Improve postinstall script

Improvements proposed here will:

1. Check if the instance is EC2 and if yes query for public hostname
1. If gitlab.rb already exists, read the external url set there and print that in the postinst message

Postinst message also got updated a bit to make commands easier to copy and separate from comments:

```
gitlab: Thank you for installing GitLab!
gitlab: To configure and start GitLab, RUN THE FOLLOWING COMMAND:

sudo gitlab-ctl reconfigure

gitlab: GitLab should be reachable at http://ec2-54-157-26-185.compute-1.amazonaws.com
gitlab: Otherwise configure GitLab for your system by editing /etc/gitlab/gitlab.rb file
gitlab: And running reconfigure again.
gitlab:
gitlab: For a comprehensive list of configuration options please see the Omnibus GitLab readme
gitlab: https://gitlab.com/gitlab-org/omnibus-gitlab/blob/master/README.md
gitlab:
gitlab:
gitlab: If you just upgraded from GitLab 7.9 or earlier, please run the following
gitlab: command:

sudo ln -sf   /opt/gitlab/bin/gitlab-ctl   /opt/gitlab/bin/gitlab-rake   /opt/gitlab/bin/gitlab-rails   /opt/gitlab/bin/gitlab-ci-rake   /opt/gitlab/bin/gitlab-ci-rails  /usr/bin/

gitlab:

```

Fixes #520
Fixes #521

See merge request !398

Use ci url for callback.

Fixes https://dev.gitlab.org/gitlab/omnibus-gitlab/issues/23

See merge request !399

Use ci url for callback.

Fixes https://dev.gitlab.org/gitlab/omnibus-gitlab/issues/23

See merge request !399

Minor updates for nginx templates.

See merge request !397

Configuration settings for auto linking ldap users and auto sign in with a provider.

Fixes #645

See merge request !396

Configuration settings for auto linking ldap users and auto sign in with a provider.

Fixes #645

See merge request !396

Omnibus software updates

Updated omnibus-software to the latest master https://github.com/chef/omnibus-software/compare/ea7a40b61917f78493c59920a2e691d56a096691...master

Changes that are of interest to us:

1. [libxml2 bumped to version with security fixes](https://github.com/chef/omnibus-software/commit/5fd898b35e8507bacb5ba1a6a32e0fcdfd5d8aed)
1. [postgresql bumped to minor version with security fixes](https://github.com/chef/omnibus-software/commit/e17a880894914ec2b0996ad573e90943b34b2c58)
1. [No need to provide md5 for ruby 2.1.6](https://github.com/chef/omnibus-software/commit/b99ecfaf98bed9a2a89d3b6ec38cc8b7492be676#diff-b7915a1c384a6ad708ed53c14a27ab10)
1. [Upgraded omnibus-ctl version](https://github.com/chef/omnibus-software/commit/98a5a429d9d2c856d00d594fd364d11bf64b4b7d). [No big changes between versions](https://github.com/chef/omnibus-ctl/compare/0.3.4...0.3.6)
1. [Newer version of redis, no need to pin anymore](https://github.com/chef/omnibus-software/commit/592f6cf72794d2ffe8171ceb3226d6f1813fb80f)
1. [Default zlib version changed, no need to pin anymore](https://github.com/chef/omnibus-software/commit/bc225b8668fa5a391ec4aff01f8057d45e1ee37a#diff-ca7c8fe95c252a70d222233155ee8c88)

Additionally in this MR:

1. Bumped chef version to 12.4.0.rc.2
1. Bumped nginx version to 1.7.12, [contains few bugfixes](http://nginx.org/en/CHANGES-1.8)
1. Updated omnibus version. Apart from new experimental features and few gem updates nothing I could find that would influence us. [Diff here](https://github.com/chef/omnibus/compare/3600fa33325ec4fcf6a2824773cea867a93fc9b8...0abab93bb67377d20c94bc4322018e2248b4a610)

See merge request !393

Fix mv/ln instructions for cookbook hacking

With the addition of the 'package' cookbook the old trick no longer
works somehow.

See merge request !394

Try to leave 1GB RAM free of Unicorn workers

If we keep adding Unicorn workers as CPU cores go up we can run out of
memory. After this change:
- the minimum number of Unicorn worker processes is 2
- if there is enough RAM, we get ncores+1 worker processes
- assuming 250MB/Unicorn, leave 1GB RAM free for other services. (1.5GB
  RAM: max 2 Unicorns, 2GB: max 4 Unicorns, 4GB: max 12 Unicorns, etc.)

See merge request !395

If we keep adding Unicorn workers as CPU cores go up we can run out of
memory. After this change:
- the minimum number of Unicorn worker processes is 2
- if there is enough RAM, we get ncores+1 worker processes
- assuming 250MB/Unicorn, leave 1GB RAM free for other services. (1.5GB
  RAM: max 2 Unicorns, 2GB: max 4 Unicorns, 4GB: max 12 Unicorns, etc.)