ssh-monitor - terminal-based SSH monitoring tool.

2025-08-25

A terminal-based SSH monitoring tool. If you need to maintain multiple physical servers and connect via SSH, this tool will help you.

Motivation

I need to maintain more than 20 physical servers at work. I connect to them via SSH and have to access each one individually to check CPU usage, disk usage, GPU usage, etc.

This process is time-consuming and doesn't scale well. I needed a better solution.

Introducing ssh-monitor

ssh-monitor is a terminal-based tool I built to solve this exact problem. It automatically discovers SSH hosts from your config and provides real-time system metrics in an interactive terminal interface.

Installation

  1. Start SSH agent and add your keys:
ssh-agent
ssh-add ~/.ssh/id_rsa
  1. Install ssh-monitor:
curl -sSL https://raw.githubusercontent.com/tsugumi-sys/ssh-monitor/main/install.sh | bash
  1. Run it:
ssh-monitor

Current Features

  • Host Discovery: Automatically finds SSH hosts from your SSH config
  • Interactive Terminal UI: Built with Ratatui for smooth navigation
  • Metrics Tracked:
    • CPU usage and historical timeline
    • Memory utilization
    • Disk usage
    • GPU metrics (when available)

How It Works

Simply run ssh-monitor and it will:

  1. Discover hosts from your SSH config file automatically
  2. Display host list - navigate with arrow keys
  3. View detailed metrics - press Enter to see real-time charts for a specific host
  4. Monitor continuously - metrics update automatically

The interface shows CPU usage timelines, memory utilization, disk space, and GPU metrics when available.

Technical Implementation: Efficient SSH Connections

One key challenge was minimizing SSH connections. A naive approach would create a new SSH connection for each metric on each server.

With 10 servers monitored every 5 seconds, that's 120 connections per minute - a big performance issue for both clients and servers.

Instead, ssh-monitor batches multiple metrics into a single SSH session:

echo __BEGIN_cpu__
bash -c '
  if [[ "$(uname)" == "Darwin" ]]; then
    sysctl -a;
  else
    lscpu;
  fi
'
echo __END_cpu__
echo __BEGIN_mem__
bash -c 'uname -s && echo __MEM__ && (free -m || (echo __MAC__ && sysctl -n hw.memsize && vm_stat))'
echo __END_mem__
echo __BEGIN_disk__
df -Pm | tail -n +2
echo __END_disk__

The output uses delimiter tags, allowing the tool to parse and extract each metric from a single command execution. This reduces connections from 120 to just 30 connections per minute.