wake/centos-gitlab-installation-faq.md

## centos-gitlab-installation-faq.md

      
    Raw
  

              centos-gitlab-installation-faq.md
            
          
    CentOS GitLab Installation FAQ


make: *** [doc/capi/.timestamp] Killed

There are several way but I can't promise which one will work:

Upgrade server memory or remove /tmp/ruby/ruby-1.9.3-p392 and redownload & re-make. Command make may be interrupted (killed) by out of memory reason. Take this as an option if you have always received make killed message after tried all other suggestions on a low memory (less than 1G) server.
Another suggestion: here.
Upgrade doxygen. (as following)

Remove doxygen
sudo yum remove doxygen

Download & install last version manually
mkdir /tmp/doxygen && cd /tmp/doxygen
svn co https://doxygen.svn.sourceforge.net/svnroot/doxygen/trunk doxygen-svn
cd doxygen-svn
./configure
make
sudo make install

sudo: gem: command not found

Use sudo -i instead of sudo, -i acquires the root user's environment, or you can use an old way solution as below
mv /usr/bin/sudo /usr/bin/sudo.ori
vi /usr/bin/sudo

#!/bin/bash
/usr/bin/sudo.ori env PATH=$PATH "$@"

Ref:sudo: gem: command not found
yaml.rb error, it seems your ruby installation is missing psych

wget http://mirror01.idc.hinet.net/EPEL/6/i386/epel-release-6-8.noarch.rpm
rpm -iUvh epel-release-6-8.noarch.rpm

Then
yum install libyaml-devel -y

Recompile ruby.
Can't locate Time/HiRes.pm in @INC

yum install perl-Time-HiRes -y

sudo: bundle: command not found

Use absolute path to execute bundle:
whereis bundle
sudo -u gitlab -H /real/path/to/bundle ....

/usr/bin/env: ruby: No such file or directory, Could not locate Gemfile

Solution A

vi /etc/profile.d/ruby.sh

Add below:
export PATH=$PATH:/usr/local/bin:/usr/local/lib

Solution B

Sudo needs -i option to make /usr/bin/env: ruby executing correctly, but it will also change the current working directory to gitlab's home and make Could not locate Gemfile error.
But the old way solution works well:
sudo -u gitlab -H env PATH=$PATH /usr/local/bin/bundle exec rake gitlab:setup RAILS_ENV=production

Ref:Stay at same working directory when changing to sudo
Error connecting to Redis on localhost

service redis start

upstream prematurely closed connection while reading response header from upstream

vi /home/gitlab/gitlab/config/unicorn.rb

# nuke workers after 30 seconds (60 is the default)
timeout 120

About sidekiq

When you run GitLab on a low memory server (like me), sidekig is possible to take 1~2 mins to start after service gitlab start or /etc/init.d/gitlab executed. Otherwise, if you get error log sidekiq memory allocate error, try to flush (or restart) redis, it may give a chance.
About unicorn

As same as sidekiq, unicorn will always crush with error message like reaped #<Process::Status: pid 1997 SIGKILL (signal 9)> worker=1 even set timeout to 300 in unicorn.rb already if you deploy on a low memory server, I get out from crashing hell finally after upgrade vps (DigitalOcean) memory from 512 to 1024