Stress Test Tool

A comprehensive multi-phase server stress testing tool written in Python. Tests HTTP, HTTPS, TCP, and ICMP layers to identify your server's breaking point under concurrent load.

Features

• 6-phase progressive load test — from baseline (20 conns) to maximum stress (2000 conns)
• Multi-protocol — HTTP, HTTPS, TCP, and ICMP ping floods
• Mix of methods — GET, POST, HEAD, OPTIONS requests with randomized paths and user agents
• Real-time stats — response times, success/failure rates, and status code distribution
• Health assessment — automatic grading from EXCELLENT to CRITICAL
• Failure analysis — breakdown of every error type with percentages
• Percentile reporting — P50, P95, P99 response times

Requirements

• Python 3.8+
• aiohttp library

pip install aiohttp

Quick Start

Option 1: Edit the target in the script

Open stress_test.py and change the DEFAULT_TARGET variable (line 23):

DEFAULT_TARGET = "your-server.com"

Then run:

python stress_test.py

Option 2: Pass target as command-line argument

python stress_test.py your-server.com
python stress_test.py 192.168.1.1
python stress_test.py https://example.com/

The tool automatically strips https://, http://, and trailing slashes.

Test Phases

Phase	Concurrency	Duration	Type
1 — Baseline	20	10s	HTTP + HTTPS
2 — Medium	100	15s	HTTP + HTTPS
3 — Heavy	500	15s	HTTP + HTTPS
4 — TCP Flood	200 per port	10s	TCP (ports 22, 80, 443)
5 — Ping Flood	500 pings	—	ICMP
6 — Maximum	2000	20s	HTTP + HTTPS

Total: ~21,000+ requests in under 2 minutes

Understanding Results

Health Assessment Grades

Grade	Failure Rate	Meaning
EXCELLENT	<1%	Server handles load perfectly
GOOD	<5%	Minor strain, acceptable
MODERATE	5-15%	Server shows visible strain
WEAK	15-30%	Server struggles significantly
CRITICAL	>30%	Server collapses under load

Response Time Metrics

• P50 (Median) — half of requests were faster than this
• P95 — 95% of requests completed within this time
• P99 — tail latency for slowest 1% of requests
• Std Dev — consistency of response times (lower = more stable)

Example Output

[PHASE 1] BASELINE TEST (Light Load)

  [STATS] HTTP Baseline (20 conns)
  +-- Total Requests    : 345
  +-- Successes         : 345 (100.0%)
  +-- Failures          : 0 (0.0%)
  +-- Response Time:
  |   +-- Avg  : 549.4 ms
  |   +-- Min  : 94.0 ms
  |   +-- Max  : 4219.0 ms
  |   +-- P50  : 250.0 ms
  |   +-- P95  : 1859.0 ms
  |   +-- P99  : 3828.0 ms
  |   +-- Std  : 729.9 ms
  +-- Status Codes     : 403: 345

Common Issues

100% failures with ClientConnectorDNSError

The domain does not resolve via your system DNS. Try:

1. Use the server's direct IP instead of the domain
2. Switch to a public DNS (e.g., Google 8.8.8.8 or Cloudflare 1.1.1.1)
3. The domain may be behind Cloudflare proxy — check with nslookup

"403 Forbidden" on all requests

Normal — the server is behind nginx and rejecting unauthenticated paths. The tool measures response availability, not content validity. A 403 is still a successful server response.

High failure rate at 500+ concurrency

Your server has a connection limit. Recommendations:

1. Increase nginx worker_connections
2. Enable keepalive_timeout and keepalive_requests
3. Deploy a CDN (Cloudflare, Fastly)
4. Upgrade VPS resources (CPU, RAM, network)

Configuration

You can customize these values inside the script:

Variable	Default	Description
DEFAULT_TARGET	"169.58.148.152"	Target IP or domain
PORTS	{80, 443, 22}	Ports to test
PATHS	8 common paths	URL paths to randomize
METHODS	GET, POST, HEAD, OPTIONS	HTTP methods to randomize
USER_AGENTS	5 agents	User-Agent strings

Adjust concurrency and duration values in each phase function call within main() to customize test intensity.

License

MIT