In 5c66f954 (Allow to configure gitlab socket just by path) we added
gitlab_socket option to allow users to configure path to Unicorn socket
just by path.

But @jacobvosmaer asks for even more - make http+unix:// stuff internal
and not even try to confuse users with e.g. description of URL-encoding
socket path and possibility to configure sockets with gitlab_url and
just direct unconditionally to gitlab_socket for that case.

So be it - for users it is now shown that the only way to configure
sockets is via gitlab_socket option. Internally we still use
http+unix:// scheme and handle it even if it is given in gitlab_url.

@jacobvosmaer noted that from user perspective it is not very handy to
have URL-encoded paths for sockets, and this way it is also not
consistent with everything else in GitLab where we use just
unix://path/to/socket instead of unix://path%2Fto%2Fsocket .

Be friends to users and allow them to configure socket path just with paths
with secondary gitlab_socket option, handler for which just encodes the
path and adds http+unix:// prefix, so internal handling is always by URL.

To protect from inconsistent input, configuration handler checks that
only one of gitlab_url or gitlab_socket can be given at a time.

NOTE CGI::escape() is used instead of URL::escape() because the latter
does not escape '/' characters by default:

    http://stackoverflow.com/questions/14989581/ruby-1-9-3-add-unsafe-characters-to-uri-escape

It is well known that UNIX sockets are faster than TCP over loopback.

E.g. on my machine according to lmbench[1] they have ~ 2 times
lower latency and ~ 2-3 times more throughput compared to TCP over
loopback:

    *Local* Communication latencies in microseconds - smaller is better
    ---------------------------------------------------------------------
    Host                 OS 2p/0K  Pipe AF     UDP  RPC/   TCP  RPC/ TCP
                            ctxsw       UNIX         UDP         TCP conn
    --------- ------------- ----- ----- ---- ----- ----- ----- ----- ----
    teco      Linux 4.2.0-1  13.8  29.2 26.8  45.0  47.9  48.5  55.5  45.

    *Local* Communication bandwidths in MB/s - bigger is better
    -----------------------------------------------------------------------------
    Host                OS  Pipe AF    TCP  File   Mmap  Bcopy  Bcopy  Mem   Mem
                                 UNIX      reread reread (libc) (hand) read write
    --------- ------------- ---- ---- ---- ------ ------ ------ ------ ---- -----
    teco      Linux 4.2.0-1 1084 4353 1493 2329.1 3720.7 1613.8 1109.2 3402 1404.

The same ratio usually holds for servers.

Also UNIX sockets, since they reside on filesystem, besides being faster with
less latency, have one another nice property: access permissions to them are
managed the same way access to files is.

Because of lower latencies and higher throughput - for performance reasons, and
for easier security, it makes sense to interconnect services on one machine via
UNIX sockets and talk via TCP only to outside world.

All internal services inside GitLab can talk to each other via UNIX socket
already and only gitlab-shell was missing support to talk to Unicorn via UNIX
socket.

Let's teach gitlab-shell to talk via UNIX sockets.

[1] http://www.bitmover.com/lmbench/

~~~~

In this patch we

- add URI::HTTPUNIX to handle http+unix:// URI scheme
- add Net::HTTPUNIX to handle "connect via unix socket and then talk http"
- adjust GitlabNet#http_client_for() accordingly
- adjust documentation in config.yml.example

The http+unix:// scheme is not reinvented anew: the idea about its structure is
quite logical an was already established at least in requests-unixsocket python
package:

    http://fixall.online/theres-no-need-to-reinvent-the-wheelhttpsgithubcommsabramorequests-unixsocketurl/241810/
    https://github.com/msabramo/requests-unixsocket

Adapt gitlab-ci.yml for shared runners

After merging this we should disconnect the specific runner on gitlab-org/gitlab-shell.

See merge request !27

Version bump 2.6.6



See merge request !22

Skip 'git annex init' when using 'gcryptsetup'

'gcryptsetup' is a special git-annex feature that does its own
initialization.

See merge request !18

change images to svg

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

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

When cleaning this environment variable can be problems with the processing of non-ASCII data

Handle broken symlinks in create-hooks

If a repository contained a broken symlink named 'hooks', this would
raise ENOENT in lib/gitlab_projects.rb, which got ignored in
bin/create-hooks. This commit fixes that by making sure we handle
broken symlinks in lib/gitlab_projects.rb.

See merge request !19

If a repository contained a broken symlink named 'hooks', this would
raise ENOENT in lib/gitlab_projects.rb, which got ignored in
bin/create-hooks. This commit fixes that by making sure we handle
broken symlinks in lib/gitlab_projects.rb.

'gcryptsetup' is a special git-annex feature that does its own
initialization.

Version bump 2.6.4



See merge request !17

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

Remove keys from authorized_keys in-place

This will speed up the rm-key operation. The downside is that
authorized_keys will not shrink when you remove a key. If this ever
becomes a problem it can be fixed by running 'rake gitlab:shell:setup'.

See merge request !66

This will speed up the rm-key operation. The downside is that
authorized_keys will not shrink when you remove a key. If this ever
becomes a problem it can be fixed by running 'rake gitlab:shell:setup'.

Add git-lfs-authenticate to command white list

Add gitlab-ci yml script
Signed-off-by: Dmitriy Zaporozhets <dmitriy.zaporozhets@gmail.com>

cc @vsizov

See merge request !14

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