When your GitLab instance becomes sluggish with only 4 projects, the top output reveals the real culprit:

PID USER      PR  NI    VIRT    RES    SHR S  %CPU %MEM     TIME+ COMMAND
8760 git       20   0  648908 412768  14700 S   0.7 20.2   0:30.58 bundle
8799 git       20   0  513748 302632  14300 S   0.0 14.8   0:20.02 bundle

Multiple Ruby bundle processes are consuming over 60% of your 2GB memory. The process tree confirms these are GitLab components:

systemd─┬─agetty
        ├─bundle─┬─3*[bundle───{ruby-timer-thr}]
        │        └─{ruby-timer-thr}

Before diving deeper, run these to gather more evidence:

# Check GitLab service status
sudo gitlab-ctl status

# View memory usage breakdown
sudo gitlab-psql -c "SELECT pg_stat_activity.pid, pg_stat_activity.query_start, 
pg_stat_activity.state, pg_stat_activity.query FROM pg_stat_activity;"

# Check Unicorn workers
sudo grep worker_processes /var/opt/gitlab/gitlab-rails/etc/unicorn.rb

For small servers (2-4GB RAM), try these configuration tweaks in /etc/gitlab/gitlab.rb:

# Reduce Unicorn workers
unicorn['worker_processes'] = 2

# Limit Sidekiq concurrency
sidekiq['concurrency'] = 5

# Adjust PostgreSQL settings
postgresql['shared_buffers'] = "256MB"
postgresql['work_mem'] = "4MB"

# Enable swap as emergency measure (if not already present)
swap_file '/var/opt/gitlab/swap' do
  size 1024
end

Apply changes with:

sudo gitlab-ctl reconfigure

If problems persist after configuration changes, consider these advanced troubleshooting steps:

# Install memory_profiler gem
sudo /opt/gitlab/embedded/bin/gem install memory_profiler

# Create diagnostic script at /tmp/memcheck.rb
require 'memory_profiler'
report = Memory_profiler.report do
  # Load problematic GitLab components
  require_relative '/opt/gitlab/embedded/service/gitlab-rails/config/environment'
end
report.pretty_print

For Ubuntu 16.04 systems, consider these upgrade options:

# Option 1: Upgrade OS first
sudo do-release-upgrade

# Option 2: Migrate to Omnibus package
curl https://packages.gitlab.com/install/repositories/gitlab/gitlab-ce/script.deb.sh | sudo bash
sudo apt-get install gitlab-ce

Remember to backup first:

sudo gitlab-rake gitlab:backup:create

When running Gitlab CE on a Ubuntu 16.04 server with limited resources (2GB RAM in this case), we observed severe performance degradation during operations like git pushes and web interface access. The top output revealed multiple bundle processes consuming over 60% of total memory collectively:

PID USER      PR  NI    VIRT    RES    SHR S  %CPU %MEM     TIME+ COMMAND
8760 git       20   0  648908 412768  14700 S   0.7 20.2   0:30.58 bundle
8799 git       20   0  513748 302632  14300 S   0.0 14.8   0:20.02 bundle

These processes are part of Gitlab's Ruby on Rails architecture. The pstree output shows they're worker processes for handling requests:

bundle─┬─3*[bundle───{ruby-timer-thr}]
       └─{ruby-timer-thr}

First, let's verify and adjust the Unicorn worker configuration (Gitlab's application server):

# Edit the Gitlab configuration
sudo nano /etc/gitlab/gitlab.rb

# Recommended settings for 2GB server
unicorn['worker_processes'] = 2
unicorn['worker_memory_limit_min'] = "250MB"
unicorn['worker_memory_limit_max'] = "300MB"

# Apply changes
sudo gitlab-ctl reconfigure

For systems with memory constraints, consider these additional optimizations:

# Reduce Sidekiq concurrency (background jobs)
sidekiq['concurrency'] = 5

# Adjust PostgreSQL shared buffers
postgresql['shared_buffers'] = "256MB"

# Enable memory killer
unicorn['oom_score'] = -900

# Apply changes
sudo gitlab-ctl reconfigure
sudo systemctl restart gitlab-runsvdir

Install these utilities for deeper memory analysis:

sudo apt-get install -y ruby-dev ruby-bundler
sudo gem install derailed_benchmarks

# Run memory profiler
bundle exec derailed bundle:mem

For consistently overloaded systems:

• Upgrade to at least 4GB RAM (Gitlab's minimum recommended)
• Consider switching to Puma instead of Unicorn (Gitlab 13.0+)
• Implement swap space as temporary relief (not ideal for production)

Regularly monitor memory usage with this cron job:

# Add to /etc/crontab
0 * * * * root /usr/bin/curl -s http://localhost/-/metrics | grep ruby_process_resident_memory_bytes