30secs

30secs is an ultra-lightweight system monitoring tool that takes system snapshots every 30 seconds (default), or returns snapshots on-demand via CLI.

Features

• Real-time monitoring: CPU, memory, disk, network, and process metrics
• Multiple output formats: JSON, human-readable table, Prometheus metrics
• Alert system: Configurable threshold-based alerts
• Graceful shutdown: Handles SIGINT/SIGTERM signals
• Production-ready: Structured logging, error handling, modular architecture

⚠️ Python module names cannot start with a number, so while the project is named 30secs, the import package is thirtysecs.

Installation

Binary (No Python Required) - Recommended for K8s Nodes

Download the pre-built binary for your platform:

# Linux AMD64 (most K8s nodes)
curl -L https://github.com/<your-repo>/30secs/releases/latest/download/30secs-linux-amd64 -o 30secs

# Linux ARM64 (Graviton, etc.)
curl -L https://github.com/<your-repo>/30secs/releases/latest/download/30secs-linux-arm64 -o 30secs

# macOS Apple Silicon
curl -L https://github.com/<your-repo>/30secs/releases/latest/download/30secs-darwin-arm64 -o 30secs

# macOS Intel
curl -L https://github.com/<your-repo>/30secs/releases/latest/download/30secs-darwin-amd64 -o 30secs

# Make executable and run
chmod +x 30secs
./30secs watch -f table --alerts

One-liner for K8s Node Debugging

curl -sL https://github.com/<your-repo>/30secs/releases/latest/download/30secs-linux-amd64 -o /tmp/30secs \
  && chmod +x /tmp/30secs \
  && /tmp/30secs watch -f table --alerts -i 10

pip (PyPI) - Recommended for environments without GitHub access

# Install from PyPI (Python 3.12+ required)
pip install 30secs

# Or use pipx for isolated install
pipx install 30secs

# Or use uv
uv tool install 30secs

# Then run
30secs watch -f table --alerts -i 10

One-liner for K8s Node (pip)

pip install 30secs && 30secs watch -f table --alerts -i 10

From Source (uv)

Install uv (official docs)

curl -LsSf https://astral.sh/uv/install.sh | sh

Install/sync dependencies

uv sync

Usage

Single snapshot (JSON output)

uv run 30secs snapshot

Human-readable table output

uv run 30secs snapshot --format table

Prometheus metrics format

uv run 30secs snapshot --format prometheus

Continuous monitoring (every 30 seconds)

uv run 30secs watch

Custom interval (every 10 seconds)

uv run 30secs watch --interval 10

Watch with table format (auto-refresh)

uv run 30secs watch --format table --interval 5

Quick snapshot (without processes - faster)

uv run 30secs quick

Enable alerts

uv run 30secs watch --alerts

Save output to file

uv run 30secs watch --output metrics.jsonl

Limit number of snapshots

uv run 30secs watch --count 10 --interval 5

Exclude specific metrics

uv run 30secs snapshot --no-processes --no-network

Memory leak report for a process (table)

uv run 30secs leak <PID> --interval 2 --count 30

Memory leak report as JSON

uv run 30secs leak <PID> --format json --interval 1 --count 60

Rank top leak candidates

uv run 30secs leak top --interval 1 --count 20 --limit 5

Output Formats

JSON (default)

Complete system metrics in JSON format, suitable for log aggregation and processing.

Table

Human-readable format with emojis and organized sections:

• System info (hostname, OS, uptime)
• CPU (usage, cores, load average)
• Memory (used/total, available, swap)
• Disk (partitions, usage)
• Network (sent/received, connections)
• Processes (top by CPU/memory)

Prometheus

Metrics in Prometheus exposition format, ready for scraping.

Alerts

Built-in alert thresholds:

• CPU: > 90%
• Memory: > 90%, > 95% (critical)
• Swap: > 80%

When alerts are enabled (--alerts), the tool will:

• Log warnings when thresholds are exceeded
• Exit with code 1 if any alert is triggered (for snapshot command)

Development

Run tests

uv run pytest

Linting & Formatting

uv run ruff check .
uv run ruff format .

Type checking

uv run mypy src

Project Structure

src/thirtysecs/
├── __init__.py          # Package init
├── __main__.py          # Entry point
├── cli.py               # CLI interface
├── config.py            # Configuration
├── core.py              # Core snapshot logic
├── alerts.py            # Alert system
├── errors.py            # Error definitions
├── logging.py           # Structured logging
├── collectors/          # Metric collectors
│   ├── base.py          # Base collector interface
│   ├── cpu.py           # CPU metrics
│   ├── memory.py        # Memory metrics
│   ├── disk.py          # Disk metrics
│   ├── network.py       # Network metrics
│   ├── process.py       # Process metrics
│   └── system.py        # System info
└── formatters/          # Output formatters
    ├── base.py          # Base formatter interface
    ├── json_fmt.py      # JSON output
    ├── table.py         # Table output
    └── prometheus.py    # Prometheus metrics

Environment Variables

Memory Leak Detection

30secs includes automatic memory leak detection when using --alerts:

# Monitor with leak detection (default: 10 samples, 5% growth threshold)
30secs watch -f table --alerts -i 10

# Custom thresholds via environment variables
MEMORY_LEAK_WINDOW_SIZE=20 MEMORY_LEAK_GROWTH_THRESHOLD=3.0 30secs watch --alerts -i 5

The detector tracks memory usage over a sliding window and alerts when:

• Memory growth exceeds the threshold (default: 5%)
• At least 60% of samples show an increasing trend

uv Cheat Sheet

License

MIT