Removal of gitlab.socket should be in stop function too. Fix #4313

Created by: axilleas

As @wojdec suggested in #4313 (closed), this fix first checks if gitlab.socket doesn't exist and iff the check returns a non-zero status (which means it exists), it then removes it.

I also removed the & in the restart function as I experienced some weird behaviour when I restarted the service (sidekiq didn't start after it stopped).

Tested on Debian 7.0, GitLab version 5.3.0.beta1.

Created by: coveralls

Coverage decreased (-0%) when pulling a4048012a3c1fb90bfd68216438a5e44213e8478 on axilleas:initd_script into 2e839cee on gitlabhq:master.

By Administrator on 2013-06-17T09:53:31 (imported from GitLab project)

By Administrator on 2013-06-17T09:53:31 (imported from GitLab)

Created by: dzaporozhets

nice. Can you make im mergeable again please? Mention me when ready

By Administrator on 2013-07-11T16:57:36 (imported from GitLab project)

By Administrator on 2013-07-11T16:57:36 (imported from GitLab)

Created by: coveralls

Coverage remained the same when pulling ec38d629 on axilleas:initd_script into 8cdfa86b on gitlabhq:master.

By Administrator on 2013-07-12T09:19:31 (imported from GitLab project)

By Administrator on 2013-07-12T09:19:31 (imported from GitLab)

Created by: dzaporozhets

thank you

By Administrator on 2013-07-12T12:45:26 (imported from GitLab project)

By Administrator on 2013-07-12T12:45:26 (imported from GitLab)

Created by: axilleas

no problem :)

By Administrator on 2013-07-12T17:35:06 (imported from GitLab project)

By Administrator on 2013-07-12T17:35:06 (imported from GitLab)