qemu-riscv-cluster

Project structure

-- This repository contains gRPC client and scripts to build and deploy workers in the RISC-V VM.

-- Scripts and Yocto Layers to build VM image with all dependencies

-- Sources to build and run gRPC server and worker nodes

Contribution

Recommendations for working with the repository

Setting up a pre-commit hook for formatting code using Black

• The black configuration settings are located in .pre-commit-config.yaml

• After cloning the project, download all the necessary dependencies using pip install -r requirements.txt in a virtual environment. Run the pre-commit install command in the virtual environment once. Next, the hook will be triggered with each commit.

• If black finds a problem in the code, he will fix it, and you will need to repeat git add . and git commit -m "name_commit"

• You can check for a hook with cat command.git/hooks/pre-commit if it returns the code then the hook is installed.

• Additional information is described in the file using_black.md

Project Launch Guide

1. Clone the Repositories

# Clone the main repository
git clone https://github.com/moevm/qemu-riscv-cluster.git
cd qemu-riscv-cluster

# Initialize and update submodules
git submodule init
git submodule update

2. Dependencies for Running on Host

System Dependencies

# Install required packages
apt-get install -y \
    docker.io \
    docker-compose \
    python3 \
    python3-pip

docker plugin install grafana/loki-docker-driver:latest --alias loki --grant-all-permissions
docker plugin enable loki
sudo systemctl restart docker

2.1. Setting up service credentials

Setting up usernames and passwords for the Grafana web interface

1. Open the log_and_metric/.env file in a text editor.

2. Set the desired usernames and passwords. For example:

GF_ADMIN_USER=user
GF_ADMIN_PASSWORD=your_password

Note: If you do not set these variables, the default credentials will be used

3. Save the file.

3. Project Management

The project management script supports the following commands:

# Update submodules
./deploy_service.sh update

# Start services (default: 5 replicas)
./deploy_service.sh start [number_of_replicas]

# Stop services
./deploy_service.sh stop

# Restart services
./deploy_service.sh restart [number_of_replicas]

# Show help
./deploy_service.sh help

The script will launch:

• Grafana
• Prometheus
• Loki
• gRPC server
• Docker Compose with the controller
• Worker containers (specified number of replicas)

4. Running Tests

Test Preparation

1. Make sure all system components are up and running

2. Verify the availability of all services:

   • Grafana (port 3000)
   • Prometheus (port 9090)
   • Loki (port 3100)
   • gRPC server

Load Test Description

The load test validates the performance of the load distribution system and includes:

1. File upload testing:

   • Generating test files of a specified size and type
   • Sending files through the gRPC server
   • Validating file uploads and correctness

2. Load distribution testing:

   • Simulating multiple concurrent users
   • Checking task distribution balance across workers
   • Monitoring individual worker load

3. Performance metrics collection:

   • System response time
   • Number of processed requests
   • Size of processed files
   • Number of errors

Running the Load Test

# Run the test with default parameters
./run_load_test.sh

# Or with custom parameters
./run_load_test.sh -d 60 -c 3 -s 10KB -t text

Test parameters:

• -d, --duration – duration of the test in seconds (default: 60)
• -c, --concurrent – number of concurrent users (default: 3)
• -s, --size – size of the test file (default: 10KB)
• -t, --type – type of the test file (text or binary)

Test Results

After running the test, you'll see:

• Total number of requests
• Average response time
• 95th percentile response time
• 99th percentile response time
• Total test duration

Metrics are also available in:

• Prometheus (port 9090)
• Grafana dashboards (port 3000)
• Loki logs (port 3100)

5. Monitoring

Once all components are running, you can access:

• Grafana: http://localhost:3000
• Prometheus: http://localhost:9090
• Loki: http://localhost:3100

6. Stopping the Project

./deploy_service.sh stop