SMTPBench

A robust SMTP load testing and benchmarking tool with MX failover support and detailed logging capabilities.

📦 Available on PyPI • 🚀 Quick Start Guide • 📖 Documentation

⚠️ Important Notice

This tool will attempt to send real emails when run against any hostname. Please be aware:

• Permission Required: Ensure you have permission on both the sending and receiving networks to use this tool. Unauthorized use may violate terms of service or laws.
• Email Tracking: All emails sent by this tool include message counts, thread IDs, and unique run identifiers (UUIDs) in the subject line and body. This enables tracking and identification if issues arise.
• ISP Blocking: If running against public hostnames, your ISP may already be blocking outbound SMTP traffic (ports 25, 587, 465). You can try alternate ports, but if these ports are open, you could still be blocked or flagged.
• Not for Residential Use: This tool is not meant to be run from residential internet connections or personal systems without proper authorization.
• Not for Malicious Use: This is NOT a tool for denial of service attacks or any malicious activity.
• Intended Purpose: This is purely an email benchmarking and load testing tool designed for authorized testing of SMTP infrastructure in controlled environments.

Use responsibly and ethically.

Features

• 🚀 Multi-threaded Load Testing - Simulate concurrent SMTP connections; threads= workers × messages= each, or messages=0 to run until interrupted
• 🔄 MX Failover - Automatic MX record lookup, tried in priority order with failover to backup servers — or skip DNS entirely with lb_host=
• ✅ Pre-flight Banner Check - Verifies the first host answers before sending anything, so an unreachable server fails fast instead of after N threads of errors
• ♻️ Retry with Backoff - max_retries= attempts spaced by retry_delay=, counted and reported separately from final per-message outcomes
• ⏱️ Timeout and Pacing Control - transaction_timeout= per SMTP transaction, plus a fixed delay= or random_delay= between messages
• 🎚️ Whole-run Rate Cap - rate= messages/sec enforced by a token bucket shared across all threads
• 📊 Real-time Progress - Live progress bar with success rate metrics
• 🎨 Color-coded Output - Success rate colored by threshold: green at ≥90%, yellow at ≥70% and <90%, red below 70%
• 📈 Summary Artifact - Machine-readable summary_*.json per run with totals, latency percentiles, and per-MX counts
• 📝 Detailed Logging - Structured JSON logs for success, failures, retries, and debug info, written to logfile_output= (default ./logs)
• 🏷️ Traceable Messages - Every message carries X-SMTPBench-Run-UUID, X-SMTPBench-Thread-ID, and X-SMTPBench-Message-ID headers, so delivered mail can be tied back to the exact run and worker that sent it
• 🔒 TLS Transport Modes - starttls, implicit ssl (port 465), or none
• 🔑 SMTP AUTH - Credentials from CLI, environment, or an auto-discovered .env file (dotenv_path= to override); never written to logs or the summary
• 🖥️ Custom HELO/EHLO - Announce any hostname with client_hostname= instead of the system default
• 📎 Attachments - A static file, synthetic payloads of a fixed size or range, or random sampling from a corpus directory — with the estimated total payload printed before sending
• 📄 Body-text Corpus - Prefix each message with a randomly selected text file for realistic content
• 📇 Address Lists - Draw recipient / from / journal addresses from files, randomly or round-robin
• 💾 Offline EML Mode - Compose to .eml files on disk instead of sending, for message-format work without a mail server; content-addressed {sha256}.eml names mean identical messages deduplicate to one file
• 📬 Journal Mode - journal=true adds a second envelope recipient (journal_address=, defaulting to the recipient) to exercise compliance-journaling paths — a bcc-style copy, so it never appears in the message headers
• 🐛 Debug Mode - Detailed debugging for troubleshooting, including the SMTP wire conversation — muted around the AUTH exchange so the SASL handshake never reaches your terminal
• 🐳 Docker Support - Run in containers

Every option behind these is listed in Configuration Options.

Installation

From PyPI (Recommended)

pip install smtpbench

Upgrade to Latest Version

pip install --upgrade smtpbench

Or specify a version:

pip install smtpbench==1.2.1

From Source

git clone https://github.com/SMTPBench/SMTPBench.git
cd SMTPBench
pip install -e .

Using Docker

# Build the image. The -f is required: the file is named `dockerfile` (lowercase),
# which Docker will not find by default on case-sensitive filesystems.
docker build -f dockerfile -t smtpbench .

# Run with log volume mount
docker run --rm -v $(pwd)/logs:/app/logs smtpbench \
    recipient=test@local.lets.qa \
    port=587 \
    threads=5 \
    messages=10

# Or run without volume mount (logs stay in container)
docker run --rm smtpbench \
    recipient=test@local.lets.qa \
    port=587 \
    threads=5 \
    messages=10

Quick Start

Getting Help

# Show version
smtpbench --version
smtpbench -v

# Show full help with all options
smtpbench --help

# Also works with
smtpbench -h
smtpbench help
smtpbench ?

Basic Usage

smtpbench recipient=test@local.lets.qa port=587 threads=5 messages=10

With TLS and Custom Settings

smtpbench \
    recipient=test@local.lets.qa \
    port=587 \
    from_address=loadtest@local.lets.qa \
    threads=10 \
    messages=100 \
    tls_mode=starttls \
    retry_delay=5 \
    max_retries=3 \
    transaction_timeout=30

Using Load Balancer Instead of MX Lookup

smtpbench \
    recipient=test@local.lets.qa \
    lb_host=smtp.local.lets.qa \
    port=587 \
    threads=5 \
    messages=20

With Attachments

Attach a static file to every generated message:

smtpbench \
    recipient=test@local.lets.qa \
    port=587 \
    threads=2 \
    messages=5 \
    attachment_path=./sample.pdf

Generate synthetic attachments for size-based benchmarking:

smtpbench \
    recipient=test@local.lets.qa \
    port=587 \
    threads=2 \
    messages=5 \
    attachment_size=1MB \
    attachment_count=2 \
    attachment_filename=payload.bin \
    attachment_mime_type=application/octet-stream

Attachment safety: Attachments multiply outbound traffic and downstream storage pressure. SMTPBench prints the per-message attachment size and estimated total attachment payload before sending when attachments are enabled.

As a Python Module

python -m smtpbench recipient=test@local.lets.qa port=587 threads=5 messages=10

Configuration Options

The tables below are the complete reference — every option SMTPBench accepts appears in one of them. For terminal-formatted usage with worked examples, run:

smtpbench --help

Note: --help is a quick reference, not an exhaustive one — it currently omits attachment_dir= and attachment_probability=. These tables are authoritative.

Required Parameters

Parameter	Description	Example
recipient	Target email address — or supply recipient_file= instead (see Address Lists from a File)	test@local.lets.qa
port	SMTP port number — not required when eml_out_dir= is set, since an offline run never connects (see Offline EML Output)	587 or 25
threads	Number of concurrent threads	10
messages	Messages per thread (0 for infinite)	100

Note: If any required parameter is missing, SMTPBench will display a helpful error message with examples.

Optional Parameters

Parameter	Default	Description
lb_host	(auto MX lookup)	Load balancer/SMTP host (skips MX lookup)
from_address	no-reply@localhost	Sender email address
retry_delay	20	Seconds to wait between retries
use_tls	(unset)	Deprecated — alias for tls_mode: true→starttls, false→none. Prints a deprecation warning. Beware that use_tls=true forces starttls even on port 465, where leaving it unset selects ssl — so it is not a no-op. Use tls_mode instead.
delay	0	Fixed delay between messages (seconds)
random_delay	false	Random 1-15 second delay between messages
transaction_timeout	20	SMTP transaction timeout (seconds)
max_retries	3	Maximum retry attempts per message
client_hostname	(system hostname)	Client hostname for SMTP HELO/EHLO
logfile_output	./logs	Directory for log files
journal	false	Enable journal mode
journal_address	(same as recipient)	Email address for journal copies
recipient_file	(none)	File of recipient addresses, one per line. Requires lb_host= or eml_out_dir=. See Address Lists from a File
from_file	(none)	File of From addresses. See Address Lists from a File
journal_file	(none)	File of journal addresses. Requires journal=true. See Address Lists from a File
recipient_file_order	random	Selection order for recipient_file: random or roundrobin
from_file_order	random	Selection order for from_file: random or roundrobin
journal_file_order	random	Selection order for journal_file: random or roundrobin
debug	false	Enable debug logging
attachment_path	(none)	Attach a specific file to every message
attachment_size	(none)	Generate synthetic attachment(s); accepts a fixed size or range (512KB, 10KB-2MB)
attachment_dir	(none)	Sample attachments from a corpus directory (mutually exclusive with attachment_path/attachment_size)
attachment_probability	1.0	Probability (0.0–1.0) that any given message gets an attachment (used with attachment_size/attachment_dir)
attachment_count	1	Number of attachments per message; accepts a range (1-3)
attachment_filename	source/generated name	Override attachment filename (payload.bin becomes payload-1.bin, payload-2.bin, etc. when count > 1)
attachment_mime_type	auto-detected / application/octet-stream	Override attachment MIME type
tls_mode	starttls (or ssl on port 465)	TLS transport: starttls, ssl, or none. Use ssl for port 465 implicit-TLS. use_tls= is a deprecated alias.
rate	(none)	Whole-run cap in messages/sec, enforced by a shared token bucket. Mutually exclusive with delay=/random_delay=.
username	(none)	SMTP AUTH username. Prefer SMTPBENCH_USER env var or .env over CLI (CLI credentials are visible in ps/shell history).
password	(none)	SMTP AUTH password. Prefer SMTPBENCH_PASS env var or .env over CLI.
dotenv_path	(auto-discovered)	Path to a .env file for credential resolution. When unset, python-dotenv searches the current directory and its parents for a .env file.
body_text_dir	(none)	Prefix each message body with a randomly selected text file from this directory. Selection is deterministic per run (seeded from run UUID). Only the chosen filename and character length are logged — never the excerpt text.
eml_out_dir	(none)	Offline mode: write each composed message to this directory as {sha256}.eml instead of sending over SMTP. Skips DNS/MX lookup and banner check. Identical message bytes deduplicate to a single file. Composes with attachments and body_text_dir.

SMTP Authentication

SMTPBench resolves credentials in this order: CLI arguments → environment variables → .env file.

Preferred — set environment variables or use a .env file so credentials are not visible in process listings:

# Via environment variables
export SMTPBENCH_USER=myuser
export SMTPBENCH_PASS=mypassword
smtpbench recipient=test@example.com port=587 threads=5 messages=10

# Or via a .env file (auto-discovered in the current directory or its parents)
echo "SMTPBENCH_USER=myuser" >> .env
echo "SMTPBENCH_PASS=mypassword" >> .env
smtpbench recipient=test@example.com port=587 threads=5 messages=10

# Custom .env path
smtpbench recipient=test@example.com port=587 threads=5 messages=10 dotenv_path=/etc/smtpbench.env

Discouraged — passing credentials on the CLI makes them visible in ps, top, and shell history:

# ⚠ Credentials visible in ps/shell history — prefer env/.env instead
smtpbench recipient=test@example.com port=587 threads=5 messages=10 \
    username=myuser password=mypassword

SMTPBench prints a warning when credentials are supplied this way.

TLS Modes

Use tls_mode= to control the TLS transport. The default is starttls (upgrade an initially plain connection); port 465 defaults to ssl (implicit TLS from the start):

# STARTTLS on port 587 (default)
smtpbench recipient=test@example.com port=587 threads=5 messages=10 tls_mode=starttls

# Implicit SSL on port 465
smtpbench recipient=test@example.com port=465 threads=5 messages=10 tls_mode=ssl

# Plain (no TLS)
smtpbench recipient=test@example.com port=25 threads=5 messages=10 tls_mode=none

The use_tls=true/false flag is a deprecated alias for tls_mode=starttls/none. It is not equivalent to omitting TLS options: use_tls=true resolves to starttls unconditionally, so on port 465 it overrides the implicit-ssl default. Setting both tls_mode and use_tls is allowed — tls_mode wins and a warning is printed.

Rate Limiting

Cap the whole-run throughput with rate= (messages/sec). The limit is enforced by a shared token bucket across all threads and is mutually exclusive with delay=/random_delay=:

# Cap at 10 messages/sec across all threads
smtpbench recipient=test@example.com port=587 threads=10 messages=100 rate=10

Corpus-directory Attachments

Sample attachments randomly from a directory of real files:

smtpbench \
    recipient=test@example.com \
    port=587 \
    threads=5 \
    messages=20 \
    attachment_dir=./corpus \
    attachment_probability=0.8 \
    attachment_count=1-3

• attachment_dir= — directory to sample from (mutually exclusive with attachment_path/attachment_size)
• attachment_probability= — probability (0.0–1.0) that a given message gets attachments (default: 1.0)
• attachment_count= — number of files to attach; accepts a range like 1-3 (default: 1)
• attachment_size= also accepts a range, e.g. 10KB-2MB, to generate variable-sized synthetic attachments

Per-message selection is seeded from the run UUID so results are reproducible.

Note: attachment_filename= cannot be combined with attachment_dir=. Corpus files keep their own names — renaming every sampled file to one fixed name would defeat the point of sampling a corpus — so SMTPBench rejects the combination rather than silently ignoring one of the two.

Body-text Prefix

Prepend a randomly selected text file from a local directory above the standard message body:

smtpbench \
    recipient=test@example.com \
    port=587 \
    threads=5 \
    messages=20 \
    body_text_dir=./text-corpus

• body_text_dir=PATH — directory of .txt (or any text) files to sample from
• Selection is deterministic per run: seeded from the run UUID, so re-runs with the same UUID pick the same file
• The subject line, tracking headers, and footer are preserved unchanged
• Only the chosen filename and character length are logged — the excerpt text is never written to logs

Offline EML Output

Write composed messages to disk as .eml files instead of sending over SMTP:

smtpbench \
    recipient=test@example.com \
    threads=5 \
    messages=20 \
    eml_out_dir=./eml-output

• eml_out_dir=PATH — directory to write {sha256}.eml files into
• Fully offline: no DNS/MX lookup and no banner check are performed; the recipient is used only as a message header
• port= is not required here and has nothing to apply to. It is still accepted; if omitted, the summary records "port": null
• Identical message bytes (same subject, body, attachments) deduplicate to a single file via SHA-256 naming
• Composes correctly with attachment_* options and body_text_dir= — all composition happens before the offline fork

Address Lists from a File

Supply a plain-text file of addresses for recipient, from_address, or journal_address. SMTPBench picks one address per message, independently per field.

Parameter	Description
recipient_file=PATH	File of recipient addresses. Requires lb_host= (fixed relay) or eml_out_dir= (offline).
from_file=PATH	File of From addresses.
journal_file=PATH	File of journal addresses. Requires journal=true.
recipient_file_order=random|roundrobin	Selection order for recipient file (default: random).
from_file_order=random|roundrobin	Selection order for from file (default: random).
journal_file_order=random|roundrobin	Selection order for journal file (default: random).

Rules and constraints:

• A *_file key overrides and cannot be combined with its single-value sibling (recipient, from_address, or journal_address).
• recipient_file requires lb_host= (a fixed relay) or eml_out_dir= (offline mode). MX-based delivery with a multi-domain recipient file is not supported; all recipients are delivered through the one relay.
• journal_file requires journal=true.
• A *_file_order key requires its matching *_file key. Ordering an address list you never supplied is a mistake rather than a no-op, so SMTPBench rejects it.
• journal=true combined with recipient_file= requires an explicit journal_address= or journal_file=. Journaling normally falls back to the single recipient, and there isn't one in that combination, so SMTPBench fails fast rather than journaling nowhere.

File format:

• One address per line.
• Blank lines and lines starting with # are ignored.
• Every address is validated: must contain exactly one @ with a non-empty local part and domain.

Selection order:

• random (default) — pick a random address each message.
• roundrobin — cycle through addresses in order; gives even coverage across the run.

Logging:

• The chosen from and journal addresses are logged per message in the success/fail/retry log entries.
• The run summary records file path, address count, and order — never the raw addresses themselves.

Example:

smtpbench recipient_file=recipients.txt recipient_file_order=roundrobin \
          from_file=senders.txt \
          lb_host=smtp.example.com port=587 threads=5 messages=100

Note: Per-domain MX resolution for multi-domain recipient files is not supported. All recipients are delivered through the configured relay (lb_host=). Support for per-domain MX may be added in a future release, but it is generally slower and not relevant to relay benchmarking.

Output and Logging

SMTPBench creates structured JSON logs in the specified log directory:

Log Files

• success_TIMESTAMP_UUID.log - Successfully sent messages
• fail_TIMESTAMP_UUID.log - Failed message attempts
• retry_TIMESTAMP_UUID.log - Retry attempts
• debug_TIMESTAMP_UUID.log - Debug information (when debug=true)

Log Entry Format

{
  "timestamp": "2025-11-17T21:15:00",
  "run_uuid": "a1b2c3d4-e5f6-7890-abcd-ef1234567890",
  "client_hostname": "loadtest-server",
  "status": "success",
  "thread_id": 1,
  "message_id": 42,
  "duration_seconds": 0.523,
  "attempt": 1,
  "retry_number": null,
  "mx_host_used": "mx1.local.lets.qa",
  "recipients": ["test@local.lets.qa"],
  "attachments": [
    {
      "filename": "payload.bin",
      "size_bytes": 1048576,
      "mime_type": "application/octet-stream",
      "source": "generated"
    }
  ],
  "body_source": null,
  "from_used": "loadtest@local.lets.qa",
  "journal_used": null,
  "error": null
}

• body_source — the filename and character length of the body-text excerpt when body_text_dir= is in use, otherwise null. The excerpt text itself is never logged.
• from_used / journal_used — the addresses actually selected for this message, which matter when from_file= / journal_file= are drawing from a list.
• attachments carries metadata only; attachment bytes are never written to logs.

Summary Artifact

After every run, SMTPBench writes a summary_{timestamp}_{uuid}.json file to the log directory. It contains the run configuration, send totals, per-MX sent/failed counts, and latency percentiles:

{
  "run_uuid": "a1b2c3d4-e5f6-7890-abcd-ef1234567890",
  "client_hostname": "loadtest-server",
  "started_at": "2026-08-04_10-30-00",
  "elapsed_seconds": 42.5,
  "config": {
    "threads": 5,
    "messages": 100,
    "rate": null,
    "tls_mode": "starttls",
    "auth": false,
    "port": 587,
    "offline": false,
    "body_text_dir": false,
    "address_lists": {
      "recipient": null,
      "from": {"file": "senders.txt", "count": 250, "order": "random"},
      "journal": null
    }
  },
  "totals": {
    "sent": 495,
    "failed": 5,
    "retried": 2,
    "success_rate": 99.0
  },
  "latency_ms": {
    "p50": 120,
    "p95": 350,
    "p99": 510,
    "max": 820
  },
  "per_mx": {
    "mx1.example.com": {"sent": 495, "failed": 5}
  }
}

A few things worth knowing about these numbers:

• totals.sent and totals.failed are per-message final outcomes, so they add up to the number of messages attempted. totals.retried counts retry attempts and is reported separately — it is not part of that sum. A message that failed twice and then succeeded contributes 1 to sent and 2 to retried, not 2 to failed.
• config.auth is a boolean, not a username. Credentials never appear in the summary, the logs, or debug output.
• config.address_lists records file path, address count, and order only — never the addresses themselves.
• latency_ms is computed from successful sends only, and is null when nothing succeeded — hence the (d['latency_ms'] or {}) guard in the CI example below.

Use the summary file to gate CI pipelines on latency budgets:

# Fail the build if p95 latency exceeds 2s (checks the newest summary file)
python -c "import json,glob,os,sys; f=max(glob.glob('logs/summary_*.json'), key=os.path.getmtime); d=json.load(open(f)); sys.exit(1 if (d['latency_ms'] or {}).get('p95',0) > 2000 else 0)"

Email Message Format

Each email sent by SMTPBench includes tracking headers and identifiers for correlation:

From: loadtest@local.lets.qa
To: test@local.lets.qa
Subject: Quick test from thread 3 message 7 [a1b2c3d4-e5f6-7890-abcd-ef1234567890]
X-SMTPBench-Run-UUID: a1b2c3d4-e5f6-7890-abcd-ef1234567890
X-SMTPBench-Thread-ID: 3
X-SMTPBench-Message-ID: 7
Content-Type: multipart/mixed; boundary="===============1234567890=="
MIME-Version: 1.0

--===============1234567890==
Content-Type: text/plain; charset="us-ascii"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit

Quick test from thread 3 message 7 [a1b2c3d4-e5f6-7890-abcd-ef1234567890]

--
SMTPBench Load Testing Tool
https://github.com/SMTPBench/SMTPBench
--===============1234567890==--

Key Fields That Change Per Message:

• Subject - Contains thread ID, message ID, and run UUID
• X-SMTPBench-Thread-ID - Identifies which thread sent the message (1 to N threads)
• X-SMTPBench-Message-ID - Message number within that thread (1 to N messages)

Key Fields That Stay Constant Per Run:

• X-SMTPBench-Run-UUID - Unique identifier for the entire test run
• From - Sender address (unless changed)
• To - Recipient address (unless changed)

Example: Messages from the Same Run

Thread 1, Message 1:

Subject: Quick test from thread 1 message 1 [a1b2c3d4-e5f6-7890-abcd-ef1234567890]
X-SMTPBench-Run-UUID: a1b2c3d4-e5f6-7890-abcd-ef1234567890
X-SMTPBench-Thread-ID: 1
X-SMTPBench-Message-ID: 1

Thread 1, Message 2:

Subject: Quick test from thread 1 message 2 [a1b2c3d4-e5f6-7890-abcd-ef1234567890]
X-SMTPBench-Run-UUID: a1b2c3d4-e5f6-7890-abcd-ef1234567890  ← Same run UUID
X-SMTPBench-Thread-ID: 1                                      ← Same thread
X-SMTPBench-Message-ID: 2                                     ← Different message

Thread 3, Message 7:

Subject: Quick test from thread 3 message 7 [a1b2c3d4-e5f6-7890-abcd-ef1234567890]
X-SMTPBench-Run-UUID: a1b2c3d4-e5f6-7890-abcd-ef1234567890  ← Same run UUID
X-SMTPBench-Thread-ID: 3                                      ← Different thread
X-SMTPBench-Message-ID: 7                                     ← Different message

Example: Messages from Different Runs

First run:

Subject: Quick test from thread 1 message 1 [a1b2c3d4-e5f6-7890-abcd-ef1234567890]
X-SMTPBench-Run-UUID: a1b2c3d4-e5f6-7890-abcd-ef1234567890

Second run (different UUID):

Subject: Quick test from thread 1 message 1 [f9e8d7c6-b5a4-3210-fedc-ba9876543210]
X-SMTPBench-Run-UUID: f9e8d7c6-b5a4-3210-fedc-ba9876543210  ← Different run

Use Cases for Headers:

• Tracking: Follow individual messages across distributed mail systems
• Correlation: Match emails with JSON log entries via run UUID
• Testing: Validate message delivery and filter test data
• Debugging: Identify which test run generated specific emails
• Analysis: Aggregate metrics by run UUID or thread ID

Terminal Output

SMTPBench displays real-time progress with a success rate colored by threshold — green at ≥90%, yellow at ≥70% and <90%, red below 70% — so a degrading run is visible without reading the numbers:

[INFO] Run UUID: a1b2c3d4-e5f6-7890-abcd-ef1234567890
[INFO] Client Hostname: loadtest-server
[INFO] MX lookup for local.lets.qa:
  - mx1.local.lets.qa (priority 10)
  - mx2.local.lets.qa (priority 20)
[INFO] SMTP banner check passed for mx1.local.lets.qa:587

100%|████████████████| 500/500 [02:15<00:00, 3.70msg/s, Success=487, Fail=13, Rate=97.4%]

=== SMTP Load Test Summary ===
Run UUID: a1b2c3d4-e5f6-7890-abcd-ef1234567890
Client Hostname: loadtest-server
SMTP Hosts Tried: mx1.local.lets.qa, mx2.local.lets.qa
Total Sent: 487
Total Failed: 13
Total Retried: 8
Elapsed Time: 135.42 seconds
Logs saved in: /path/to/logs
Summary written to: /path/to/logs/summary_2026-08-04_10-30-00_a1b2c3d4-....json

In offline mode (eml_out_dir=) the host line reads SMTP Hosts Tried: (offline — wrote EML files), and no MX lookup or banner check is printed.

Use Cases (adjust samples for your needs)

Load Testing

Test SMTP server capacity and performance under concurrent load:

smtpbench recipient=test@local.lets.qa port=587 threads=50 messages=1000

Failover Testing

Verify MX failover behavior by testing multiple mail servers:

smtpbench recipient=test@local.lets.qa port=25 threads=10 messages=100

Connection Testing

Quick connectivity test with minimal load:

smtpbench recipient=test@local.lets.qa port=587 threads=1 messages=1

Sustained Load Testing

Run continuous load with random delays:

smtpbench \
    recipient=test@local.lets.qa \
    port=587 \
    threads=5 \
    messages=0 \
    random_delay=true

Examples

Test with Journal Mode

smtpbench \
    recipient=test@local.lets.qa \
    port=587 \
    threads=5 \
    messages=10 \
    journal=true \
    journal_address=archive@local.lets.qa

Debug Mode for Troubleshooting

smtpbench \
    recipient=test@local.lets.qa \
    port=587 \
    threads=1 \
    messages=1 \
    debug=true

High-Volume Load Test

smtpbench \
    recipient=test@local.lets.qa \
    port=587 \
    from_address=loadtest@local.lets.qa \
    threads=100 \
    messages=1000 \
    tls_mode=starttls \
    retry_delay=10 \
    max_retries=5 \
    transaction_timeout=30 \
    logfile_output=/var/log/smtpbench

Requirements

• Python 3.9 or higher — CI runs the unit suite on 3.9, 3.10, 3.11, 3.12, 3.13, and 3.14
• Dependencies (automatically installed):
  • dnspython>=2.0.0
  • tqdm>=4.0.0
  • colorama>=0.4.0
  • python-dotenv>=1.0.0

Development

Setup Development Environment

git clone https://github.com/SMTPBench/SMTPBench.git
cd SMTPBench
pip install -e ".[dev]"

This installs the runtime dependencies plus the dev tools (pytest, pytest-cov, and ruff).

Note: ruff is pinned to an exact version (ruff==0.16.1) so local formatting matches CI byte for byte. Formatter output changes between Ruff releases, and an unpinned install would produce spurious ruff format --check failures.

Linting and Formatting

SMTPBench uses Ruff for both linting and formatting:

# Lint
ruff check .

# Lint and auto-fix
ruff check . --fix

# Format code
ruff format .

# Check formatting without changing files (as CI does)
ruff format --check .

Running Tests

SMTPBench includes both unit tests and integration tests.

Unit Tests

Run the unit test suite:

pytest -v -m "not integration"

Integration Tests

Integration tests use Docker Compose to spin up a real SMTP server and validate end-to-end functionality:

# Run integration tests (requires Docker)
pytest -v -m "integration"

# Or use the shell script
./tests/run_integration_test.sh

The integration tests:

• Start a local test mail server using local-test-mail-server
• Run SMTPBench to send emails
• Validate emails are received in the mbox file
• Verify log files are created correctly
• Check message format and content

Run All Tests

pytest -v

CI/CD

Checks run automatically on pull requests:

• Lint - Ruff lint and format checks
• Unit tests - Fast tests without external dependencies
• Integration tests - Full end-to-end tests with Docker Compose
• Coverage - Combines the unit and integration coverage data (coverage combine) and enforces a floor; per-suite numbers are also reported for visibility
• CodeQL - Static analysis via GitHub code scanning

See .github/workflows/pytest.yml for the lint, test, and coverage jobs. CodeQL runs from GitHub's default code-scanning setup rather than a workflow file in this repository.

Coverage locally

pytest -m "not integration" --cov=smtpbench --cov-report=term-missing

Troubleshooting

Getting Help

If you're unsure about available options or syntax:

smtpbench --help  # Display full help with all options and examples

Invalid Arguments

If you see an error like Invalid argument format, ensure you're using the key=value format:

# ✗ Wrong
smtpbench --recipient test@example.com

# ✓ Correct
smtpbench recipient=test@example.com port=587 threads=5 messages=10

Missing Required Parameters

If parameters are missing, SMTPBench will show which ones are required:

$ smtpbench recipient=test@example.com
✗ Missing required parameter(s): port, threads, messages

Connection Timeouts

If you're experiencing connection timeouts, try increasing the transaction_timeout:

smtpbench recipient=test@local.lets.qa port=587 threads=5 messages=10 transaction_timeout=60

MX Lookup Failures

If MX lookup is failing, use lb_host to specify the SMTP server directly:

smtpbench recipient=test@local.lets.qa lb_host=smtp.local.lets.qa port=587 threads=5 messages=10

Debug Mode

Enable debug mode for detailed SMTP protocol information:

smtpbench recipient=test@local.lets.qa port=587 threads=1 messages=1 debug=true

Check the debug log file in your logs directory for detailed information.

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

License

This project is licensed under the MIT License - see the LICENSE file for details.

Author

Randall Morse - rmorse@lets.qa

Changelog

See CHANGELOG.md for version history.

Support

For issues, questions, or contributions, please visit the GitHub repository.