Skip to main content
Improve container and simulation performance when experiencing slowdowns.

​
Slow performance

Symptoms:
  • Container builds very slowly
  • Gazebo lags severely
  • High CPU/memory usage
Inadequate resource allocation is the most common cause of poor performance. Ensure Docker has access to sufficient CPU and memory.

​
Allocate more resources

1

Open Docker Desktop settings

Docker Desktop → Settings → Resources
2

Adjust resources

Recommended settings:
  • CPUs: 4-6 cores
  • Memory: 8-16 GB
  • Disk: 50 GB+
3

Apply changes

Click “Apply & Restart”
Resources are not limited by default on Linux. Check system availability:
# Check CPU usage
top

# Check memory usage
free -h

# Check disk I/O
iostat -x 1

​
Use software rendering

If GPU acceleration is causing issues or isn’t available: 
export LIBGL_ALWAYS_SOFTWARE=1
tb3_empty
This forces Mesa to use software rendering instead of GPU acceleration. While slower, it’s more compatible.

​
Optimize Docker on Windows

1

Enable WSL 2

  • Docker Desktop → Settings → General
  • Enable “Use WSL 2 based engine”
2

Restart Docker

Click “Apply & Restart”
WSL 2 provides significantly better performance than Hyper-V on Windows.

​
Optimize Docker on macOS

1

Enable VirtioFS

  • Docker Desktop → Settings → General
  • Enable “Use VirtioFS for file sharing”
2

Apply changes

Click “Apply & Restart”
VirtioFS provides faster file sharing between host and container.

​
Clean build cache

Stale build artifacts can slow down the workspace: 
cd /workspace/turtlebot3_ws
rm -rf build install log
colcon build --symlink-install

​
Gazebo-specific performance

For Gazebo simulation performance issues:
  1. Reduce physics update rate by modifying world files
  2. Disable shadows and anti-aliasing in Gazebo settings
  3. Use simpler world models (tb3_empty instead of tb3_house)
  4. Close RViz2 when not needed for visualization

​
Disk I/O performance

If disk operations are slow:
  1. Don’t use network drives for the workspace
  2. Disable real-time antivirus scanning for Docker directories
  3. Use SSD storage instead of HDD if possible
  4. Reduce Docker Desktop disk image size if it’s grown very large

​
Container startup performance

If the container takes a long time to start:
  1. The post-start script rebuilds the workspace on every start
  2. Consider commenting out the build step in .devcontainer/post-start.sh if you don’t need automatic rebuilds
  3. Build manually only when needed with the cb alias

Build docs developers (and LLMs) love

Get started for freeTalk to us