perf-scripts-remote

Leo Performance Testing Scripts

A comprehensive Python toolkit for running and managing leo performance testing (iperf-like tool) on local and remote machines. This suite provides tools to orchestrate performance tests with flexible configuration, result parsing, and analysis.

Overview

This project includes three main scripts:

server.py - Runs leo in server mode locally or remotely
client.py - Runs leo in client mode against a server
orchestrator.py - Manages complete test workflows with server and client coordination

Features

✅ Local and remote execution via SSH
✅ Configurable test parameters via JSON
✅ Automatic output parsing and result collection
✅ Performance metrics analysis (throughput, connections, success rates)
✅ JSON-based result export
✅ Support for TCP and UDP profiling
✅ Multiple concurrent test scenarios
✅ Comprehensive logging

Installation

Prerequisites

Python 3.7+
Access to leo binary on target machines
SSH access for remote tests (optional)

Setup

Clone or download the scripts to your project directory
Install dependencies:

pip install paramiko  # For remote SSH execution

Or install from requirements:

pip install -r requirements.txt

Quick Start

1. Generate Sample Configuration

python client.py --create-config tcp_mixed.json

This creates a configuration file like:

{
  "connections_per_second": 100,
  "duration_seconds": 30,
  "connection_profiles": [
    {
      "connection_type": "upload",
      "protocol": "tcp",
      "percentage": 50,
      "duration_seconds": 30
    },
    {
      "connection_type": "download",
      "protocol": "tcp",
      "percentage": 50,
      "duration_seconds": 30
    }
  ]
}

2. Start Leo Server

Local Server

python server.py --local --port 5201 --interval 1 -V

Remote Server

python server.py --remote 192.168.1.100 --user root --key ~/.ssh/id_rsa --port 5201 -V

3. Run Performance Test (Client)

Local Client

python client.py --host 127.0.0.1 --config tcp_mixed.json --output results.json -V

Remote Client

python client.py --host 10.0.0.1 --remote 192.168.1.50 --user root --key ~/.ssh/id_rsa \
  --config tcp_mixed.json --output results.json -V

4. Complete Orchestrated Test

Local Test (Both on same machine)

python orchestrator.py --local --server-host 127.0.0.1 --config tcp_mixed.json --output-dir ./results

Distributed Test

python orchestrator.py \
  --server-host 10.0.0.1 --server-user root --server-key ~/.ssh/id_rsa \
  --client-host 10.0.0.2 --client-user root --client-key ~/.ssh/id_rsa \
  --config tcp_mixed.json --output-dir ./results

Configuration File Format

tcp_mixed.json Example

{
  "connections_per_second": 100,
  "duration_seconds": 30,
  "connection_profiles": [
    {
      "connection_type": "upload",
      "protocol": "tcp",
      "percentage": 50,
      "duration_seconds": 30
    },
    {
      "connection_type": "download",
      "protocol": "tcp",
      "percentage": 50,
      "duration_seconds": 30
    }
  ]
}

Configuration Parameters

connections_per_second: Rate of connection creation
duration_seconds: Total test duration
connection_profiles: Array of connection scenarios
- connection_type: "upload", "download", or "bidirectional"
- protocol: "tcp" or "udp"
- percentage: Percentage of total connections
- duration_seconds: Duration for this profile

Command Reference

server.py

python server.py [OPTIONS]

Options:
  --local                    Run server locally
  --remote HOST              Run server on remote host
  --port PORT                Leo server port (default: 5201)
  --interval SECONDS         Reporting interval (default: 1)
  --user USERNAME            SSH username (default: root)
  --key KEY_FILE             SSH private key
  --leo-path PATH            Path to leo binary (default: ./leo)
  --leo-path-remote PATH     Path to leo on remote host
  --ulimit LIMIT             File descriptor limit (default: 1048576)
  --verbose, -V              Enable verbose output

client.py

python client.py [OPTIONS]

Options:
  --host HOST                Server host (REQUIRED)
  --port PORT                Server port (default: 5201)
  --client-id ID             Client identifier (default: leo_client)
  --config FILE              Configuration JSON file
  --local                    Run client locally
  --remote HOST              Run client on remote host
  --user USERNAME            SSH username (default: root)
  --key KEY_FILE             SSH private key
  --leo-path PATH            Path to leo binary (default: ./leo)
  --leo-path-remote PATH     Path to leo on remote host
  --output FILE              Save results to JSON file
  --verbose, -V              Enable verbose output
  --create-config FILE       Create sample config and exit

orchestrator.py

python orchestrator.py [OPTIONS]

Options:
  --local                    Run local server and client
  --server-host HOST         Server host (REQUIRED)
  --server-port PORT         Server port (default: 5201)
  --server-user USER         Server SSH user (default: root)
  --server-key FILE          Server SSH key
  --client-host HOST         Client host
  --client-user USER         Client SSH user (default: root)
  --client-key FILE          Client SSH key
  --client-id ID             Client identifier (default: leo_client)
  --config FILE              Configuration file (REQUIRED)
  --output-dir DIR           Results directory (default: ./results)
  --leo-path PATH            Leo binary path (default: ./leo)
  --server-wait SECONDS      Wait before starting client (default: 5)
  --test-duration SECONDS    Max test duration (default: 300)

Output and Results

Console Output

During test execution, you'll see real-time metrics:

[rms1]      1.0-46.0     sec | Connections: 3697/4482 | TCP: 3697/4482 
(UP:1553/2213 DN:2144/2269 BI:0/0 IN:0/0) | UDP: 0/0 
(UP:0/0 DN:0/0 BI:0/0 IN:0/0) | TCP TX:143.7 Mbps RX:34.5 Mbps | 
UDP TX:0.0 bps RX:0.0 bps

JSON Results File

Example results.json:

{
  "client_id": "leo_client",
  "server_host": "10.0.0.1",
  "server_port": 5201,
  "test_start": "2024-06-19T10:30:45.123456",
  "total_results": 30,
  "results": [
    {
      "timestamp": "2024-06-19T10:30:46.123456",
      "time_range": "1.0-2.0",
      "connections": "100/150",
      "tcp": {
        "completed": 100,
        "total": 150,
        "upload": "50/75",
        "download": "50/75",
        "tx_mbps": 143.7,
        "rx_mbps": 34.5
      },
      "udp": {
        "completed": 0,
        "total": 0,
        "tx_bps": 0.0,
        "rx_bps": 0.0
      }
    },
    ...
  ]
}

Summary Report

After test completion, a summary is printed:

================================================================================
Performance Test Summary for Client: leo_client
================================================================================

Test Duration: 30 seconds

TCP Throughput:
  TX: Avg=125.6 Mbps, Max=164.5 Mbps
  RX: Avg=28.3 Mbps, Max=34.5 Mbps

Connections:
  Max Connections: 3921

Final Status (at 1.0-30.0 sec):
  Total Connections: 3921/4982
  Success Rate: 78.7%
================================================================================

Typical Workflow

Scenario 1: Local Performance Testing

Start server:
```
python server.py --local -V
```

In another terminal, run test:

python client.py --host 127.0.0.1 --config tcp_mixed.json --output test_results.json -V

Scenario 2: Remote Server, Local Client

Start remote server:

python server.py --remote 10.0.0.1 --user root --key ~/.ssh/server_key -V

Run client test locally:

python client.py --host 10.0.0.1 --config tcp_mixed.json --output test_results.json -V

Scenario 3: Fully Automated with Orchestrator

python orchestrator.py \
  --server-host 10.0.0.1 --server-user root --server-key ~/.ssh/id_rsa \
  --config tcp_mixed.json \
  --output-dir ./results \
  --test-duration 60

Troubleshooting

Issue: "leo executable not found"

Solution: Ensure leo binary is in the correct path or specify with --leo-path:

python server.py --local --leo-path /usr/local/bin/leo

Issue: SSH Connection Refused

Solution: Check SSH access and key permissions:

# Verify SSH access
ssh -i ~/.ssh/id_rsa [email protected] echo "SSH OK"

# Check key permissions
chmod 600 ~/.ssh/id_rsa

Issue: Port Already in Use

Solution: Change the port:

python server.py --local --port 5202 -V
python client.py --host 127.0.0.1 --port 5202 --config tcp_mixed.json -V

Issue: "paramiko library not found"

Solution: Install paramiko for remote execution:

pip install paramiko

Issue: Too many open files

Solution: Increase system file descriptor limit before starting server:

# Adjust the ulimit
python server.py --local --ulimit 2097152 -V

Performance Tips

Increase File Descriptors: Use appropriate --ulimit value:
```
python server.py --local --ulimit 2097152 -V
```
Adjust Connections Per Second: Modify in config file:
```
"connections_per_second": 1000
```
Monitor System Resources: Run in separate terminal:
```
watch -n 1 'ps aux | grep leo'
```
Multiple Clients: Run multiple client tests against same server for load testing

Logs and Debugging

Logs are saved in the results directory:

server_TIMESTAMP.log - Server execution log
client_TIMESTAMP.log - Client execution log

Enable verbose output with -V flag for detailed logs.

Advanced Usage

Starting Multiple Clients Against One Server

Terminal 1 - Server:

python server.py --local -V

Terminal 2 - Client 1:

python client.py --host 127.0.0.1 --client-id client1 --config tcp_mixed.json -V

Terminal 3 - Client 2:

python client.py --host 127.0.0.1 --client-id client2 --config tcp_mixed.json -V

Custom Configuration Profiles

Create aggressive.json:

{
  "connections_per_second": 500,
  "duration_seconds": 60,
  "connection_profiles": [
    {
      "connection_type": "upload",
      "protocol": "tcp",
      "percentage": 100,
      "duration_seconds": 60
    }
  ]
}

Then run:

python client.py --host 10.0.0.1 --config aggressive.json -V

Dependencies

Python 3.7+: Core runtime
paramiko: SSH client library (optional, for remote execution)

License

These scripts are provided as-is for performance testing with leo.

Contributing

To enhance these scripts:

Add new metrics parsing
Support additional protocols
Implement result visualization
Add database storage support

Support

For issues or questions:

Check the Troubleshooting section
Review command help: python script.py --help
Enable verbose output with -V flag
Check log files in results directory

Name		Name	Last commit message	Last commit date
parent directory ..
utils		utils
EXAMPLES.py		EXAMPLES.py
README.md		README.md
config.yaml		config.yaml
leo_TcsSession_perf_matrix_results.xlsx		leo_TcsSession_perf_matrix_results.xlsx
perf_matrix_results.xlsx		perf_matrix_results.xlsx
requirements.txt		requirements.txt
run_perf_test.py		run_perf_test.py
server.txt		server.txt
server_port_5201.txt		server_port_5201.txt
server_port_5202.txt		server_port_5202.txt
server_port_5203.txt		server_port_5203.txt
tcp_mixed.json		tcp_mixed.json

FilesExpand file tree

perf-scripts-remote

Directory actions

More options

Directory actions

More options

Latest commit

History

perf-scripts-remote

Folders and files

parent directory

README.md

Leo Performance Testing Scripts

Overview

Features

Installation

Prerequisites

Setup

Quick Start

1. Generate Sample Configuration

2. Start Leo Server

Local Server

Remote Server

3. Run Performance Test (Client)

Local Client

Remote Client

4. Complete Orchestrated Test

Local Test (Both on same machine)

Distributed Test

Configuration File Format

tcp_mixed.json Example

Configuration Parameters

Command Reference

server.py

client.py

orchestrator.py

Output and Results

Console Output

JSON Results File

Summary Report

Typical Workflow

Scenario 1: Local Performance Testing

Scenario 2: Remote Server, Local Client

Scenario 3: Fully Automated with Orchestrator

Troubleshooting

Issue: "leo executable not found"

Issue: SSH Connection Refused

Issue: Port Already in Use

Issue: "paramiko library not found"

Issue: Too many open files

Performance Tips

Logs and Debugging

Advanced Usage

Starting Multiple Clients Against One Server

Custom Configuration Profiles

Dependencies

License

Contributing

Support