Skip to content

Testing Network Automation Scripts

Why Testing Network Automation Matters

You've built automation that configures devices, validates states, and makes changes at scale.

But here's the problem: You can't test in production.

Running your automation configure_vlan() function without tests means:

  • Bugs discovered after devices are already changed
  • Configuration rolled back manually
  • Customer impact, firefighting, blame
  • Lost trust in automation

Testing catches mistakes before they reach production.

The Business Case: Why Test Automation Code?

The Reality

Your automation code is critical infrastructure code. It:

  • Makes changes to live production networks
  • Affects hundreds of devices simultaneously
  • Has compliance and audit implications
  • Is expensive to fix when it fails

Yet many teams test automation by:

  • ❌ Running it on test devices once
  • ❌ Hoping it works when run on production
  • ❌ Discovering bugs in production via customer complaints

The Cost of Not Testing

Scenario Cost
Bug caught in unit test 5 minutes to fix
Bug caught before deployment 15 minutes to diagnose
Bug caught in production Hours of downtime + customer impact + reputation damage

The Solution: Comprehensive Testing

# Testing lets you verify:
# ✅ Decorator logic (retry, rate limiting, audit logging)
# ✅ Business logic (config generation, validation)
# ✅ Error handling (what happens when device fails?)
# ✅ Edge cases (empty config list, timeouts)
# ✅ Integration (decorators + business logic together)
# ✅ All of this WITHOUT touching real devices

Testing Fundamentals

Test Types & Their Purpose

Type Purpose Speed Real Devices?
Unit Tests Individual functions in isolation ~0.1s No (mocked)
Integration Tests Multiple components together ~1s Mocked devices
End-to-End Tests Full workflow (optional) ~30s+ Real lab devices
Property Tests Functions work for ANY valid input ~1s No (generated inputs)

For production automation: Focus on unit + integration tests. E2E tests require lab setup and aren't necessary for most cases.

Pytest Setup

Install pytest:

pip install pytest pytest-mock pytest-cov

Project structure:

my_network_automation/
├── src/
│   ├── __init__.py
│   ├── decorators.py      # Retry, logging, monitoring decorators
│   ├── device_ops.py      # Device operations (configure, verify)
│   └── utils.py           # Helper functions
├── tests/
│   ├── __init__.py
│   ├── test_decorators.py
│   ├── test_device_ops.py
│   ├── test_utils.py
│   └── conftest.py        # Shared fixtures
└── pytest.ini

Pattern 1: Testing Decorators

The Problem

Your decorators handle critical concerns (retry, logging, rate limiting). If a decorator breaks, everything breaks.

Testing @retry Decorator

# src/decorators.py
import functools
import logging
import time

logger = logging.getLogger(__name__)

def retry(max_attempts=3, delay=1, backoff=2, exceptions=(Exception,)):
    """Retry decorator with exponential backoff."""
    def decorator(func):
        @functools.wraps(func)
        def wrapper(*args, **kwargs):
            current_delay = delay
            for attempt in range(1, max_attempts + 1):
                try:
                    return func(*args, **kwargs)
                except exceptions as e:
                    if attempt == max_attempts:
                        raise
                    time.sleep(current_delay)
                    current_delay *= backoff
        return wrapper
    return decorator

def log_audit(action):
    """Log before and after an audited operation."""
    def decorator(func):
        @functools.wraps(func)
        def wrapper(*args, **kwargs):
            logger.info("Starting action", extra={"action": action})
            result = func(*args, **kwargs)
            logger.info("Completed action", extra={"action": action})
            return result
        return wrapper
    return decorator
# tests/test_decorators.py
import pytest
import time
from unittest.mock import MagicMock, patch
from src.decorators import retry

class TestRetryDecorator:
    """Test retry decorator behaviour."""

    def test_succeeds_on_first_attempt(self):
        """Function succeeds immediately—no retries needed."""
        @retry(max_attempts=3)
        def succeeds():
            return "success"

        result = succeeds()
        assert result == "success"

    def test_retries_on_transient_failure(self):
        """Function fails twice, succeeds on third—should return success."""
        attempts = []

        @retry(max_attempts=3, delay=0)  # delay=0 speeds up test
        def fails_twice():
            attempts.append(1)
            if len(attempts) < 3:
                raise ConnectionError("Transient failure")
            return "success"

        result = fails_twice()

        assert result == "success"
        assert len(attempts) == 3  # Verify it actually retried

    def test_raises_after_max_attempts(self):
        """After max attempts, exception should be raised."""
        @retry(max_attempts=2, delay=0)
        def always_fails():
            raise ConnectionError("Persistent failure")

        with pytest.raises(ConnectionError, match="Persistent failure"):
            always_fails()

    def test_exponential_backoff(self):
        """Verify backoff multiplier is applied correctly."""
        attempts = []

        @retry(max_attempts=3, delay=0, backoff=2)
        def fails_twice():
            attempts.append(time.time() if attempts else 0)
            if len(attempts) < 3:
                raise ConnectionError()
            return "success"

        # For precise timing tests, use @patch to mock time.sleep
        with patch('time.sleep') as mock_sleep:
            @retry(max_attempts=3, delay=1, backoff=2)
            def fails_twice_with_sleep():
                if mock_sleep.call_count < 2:
                    raise ConnectionError()
                return "success"

            fails_twice_with_sleep()

            # Verify sleep called with correct delays: 1s, then 2s
            assert mock_sleep.call_count == 2
            assert mock_sleep.call_args_list[0][0][0] == 1  # First delay
            assert mock_sleep.call_args_list[1][0][0] == 2  # Second delay

    def test_catches_only_specified_exceptions(self):
        """Retry only catches specified exceptions, not others."""
        @retry(max_attempts=3, exceptions=(ConnectionError,))
        def raises_value_error():
            raise ValueError("Not caught by retry")

        with pytest.raises(ValueError):
            raises_value_error()

    def test_preserves_function_metadata(self):
        """@retry should preserve original function name and docstring."""
        @retry()
        def my_function():
            """My docstring."""
            pass

        assert my_function.__name__ == "my_function"
        assert "docstring" in my_function.__doc__

Why Each Test Matters

Test Why Important
First attempt success Verify decorator doesn't break successful operations
Retry on failure Verify retry logic actually works
Max attempts exhaustion Verify you eventually give up and raise
Exponential backoff Prevent thundering herd (too-fast retries)
Exception filtering Retry only transient errors, not logic bugs
Metadata preservation Debugging and tooling must work correctly

Pattern 2: Testing Business Logic with Mocks

The Problem

Your device operations use Netmiko to send commands. You can't run actual SSH in tests.

Solution: Mock the device connection.

Testing Device Operations

# src/device_ops.py
from netmiko import ConnectHandler

def configure_interface(device, interface, ip_address, netmask):
    """Configure interface with IP address."""
    commands = [
        f"interface {interface}",
        f"ip address {ip_address} {netmask}",
        "no shutdown",
        "exit"
    ]
    device.send_command("configure terminal")
    for cmd in commands:
        device.send_command(cmd)
    device.send_command("end")
    return True

def verify_interface_config(device, interface, expected_ip):
    """Verify interface has expected IP."""
    output = device.send_command(f"show ip interface brief | include {interface}")
    return expected_ip in output
# tests/test_device_ops.py
import pytest
from unittest.mock import MagicMock, call
from src.device_ops import configure_interface, verify_interface_config

class TestDeviceOperations:
    """Test device configuration without real devices."""

    @pytest.fixture
    def mock_device(self):
        """Create a mock Netmiko device."""
        device = MagicMock()
        device.send_command = MagicMock(return_value="OK")
        return device

    def test_configure_interface_sends_correct_commands(self, mock_device):
        """Verify correct CLI commands are sent in correct order."""
        configure_interface(
            mock_device,
            interface="Gi0/0/1",
            ip_address="10.0.0.1",
            netmask="255.255.255.0"
        )

        # Verify commands sent in correct order
        expected_calls = [
            call("configure terminal"),
            call("interface Gi0/0/1"),
            call("ip address 10.0.0.1 255.255.255.0"),
            call("no shutdown"),
            call("exit"),
            call("end"),
        ]

        assert mock_device.send_command.call_args_list == expected_calls

    def test_configure_interface_returns_true_on_success(self, mock_device):
        """Function should return True indicating success."""
        result = configure_interface(
            mock_device,
            interface="Gi0/0/1",
            ip_address="10.0.0.1",
            netmask="255.255.255.0"
        )

        assert result is True

    def test_configure_interface_handles_device_error(self, mock_device):
        """If device fails, exception should propagate."""
        mock_device.send_command.side_effect = Exception("Device unreachable")

        with pytest.raises(Exception, match="Device unreachable"):
            configure_interface(
                mock_device,
                interface="Gi0/0/1",
                ip_address="10.0.0.1",
                netmask="255.255.255.0"
            )

    def test_verify_interface_config_detects_correct_ip(self, mock_device):
        """When interface has correct IP, should return True."""
        mock_device.send_command.return_value = "Gi0/0/1 YES manual up 10.0.0.1"

        result = verify_interface_config(
            mock_device,
            interface="Gi0/0/1",
            expected_ip="10.0.0.1"
        )

        assert result is True

    def test_verify_interface_config_detects_wrong_ip(self, mock_device):
        """When interface has different IP, should return False."""
        mock_device.send_command.return_value = "Gi0/0/1 YES manual up 192.168.1.1"

        result = verify_interface_config(
            mock_device,
            interface="Gi0/0/1",
            expected_ip="10.0.0.1"
        )

        assert result is False

Key Mocking Patterns

Pattern Use Case Example
MagicMock() Mock any object device = MagicMock()
.return_value Set what mock returns device.send_command.return_value = "output"
.side_effect Raise exception or return values sequentially device.side_effect = Exception()
.call_args_list Assert what arguments were passed assert device.call_args_list[0] == call(...)
.call_count Assert how many times called assert device.call_count == 5

Pattern 3: Testing Decorated Functions

The Problem

Your production code stacks decorators:

@retry(max_attempts=3)
@log_audit(action="configure_interface")
@perf.monitor
def configure_interface(device, interface, config):
    pass

You need to test the entire stack, not just individual decorators.

Testing Decorator Composition

# tests/test_integration.py
import pytest
from unittest.mock import MagicMock, patch
from src.decorators import retry
from src.device_ops import configure_interface

class TestDecoratorIntegration:
    """Test decorators + business logic together."""

    def test_retry_recovers_from_transient_failure(self):
        """Verify retry decorator works with real business logic."""
        mock_device = MagicMock()

        # Device fails first time, succeeds second time
        mock_device.send_command.side_effect = [
            Exception("Device timeout"),  # First call fails
            "OK",  # configure terminal
            "OK",  # interface command
            "OK",  # ip address command
            "OK",  # no shutdown
            "OK",  # exit
            "OK",  # end
        ]

        @retry(max_attempts=3, delay=0)
        def configure_with_retry(device, interface, ip):
            return configure_interface(device, interface, ip, "255.255.255.0")

        # Should succeed on second retry
        result = configure_with_retry(
            mock_device,
            interface="Gi0/0/1",
            ip="10.0.0.1"
        )

        assert result is True

    def test_retry_gives_up_after_max_attempts(self):
        """If all retries fail, exception should propagate."""
        mock_device = MagicMock()
        mock_device.send_command.side_effect = Exception("Device down")

        @retry(max_attempts=2, delay=0)
        def configure_with_retry(device, interface, ip):
            return configure_interface(device, interface, ip, "255.255.255.0")

        with pytest.raises(Exception, match="Device down"):
            configure_with_retry(
                mock_device,
                interface="Gi0/0/1",
                ip="10.0.0.1"
            )

    @patch('src.decorators.logger')
    def test_audit_logging_decorator_logs_attempts(self, mock_logger):
        """Verify audit logging captures operations."""
        from src.decorators import log_audit

        mock_device = MagicMock()
        mock_device.send_command.return_value = "OK"

        @log_audit(action="configure")
        def configure_with_logging(device, interface):
            device.send_command(f"interface {interface}")
            return True

        configure_with_logging(mock_device, "Gi0/0/1")

        # Verify something was logged (implementation details depend on your logging)
        assert mock_logger.info.called or mock_logger.debug.called

Pattern 4: Fixtures for Reusable Test Setups

The Problem

Writing mock_device = MagicMock() in every test is repetitive.

Solution: Pytest fixtures

# tests/conftest.py (shared across all tests)
import pytest
from unittest.mock import MagicMock

@pytest.fixture
def mock_device():
    """Reusable mock Netmiko device."""
    device = MagicMock()
    device.send_command = MagicMock(return_value="OK")
    return device

@pytest.fixture
def mock_device_with_output():
    """Mock device that returns realistic output."""
    device = MagicMock()
    responses = {
        "show ip route": "10.0.0.0/8 via 192.168.1.1",
        "show running-config": "hostname R1\nip domain-name example.com",
        "show interfaces brief": "Gi0/0/1 YES manual up 10.0.0.1",
    }
    device.send_command = MagicMock(side_effect=lambda command, **kwargs: responses.get(command))
    return device

@pytest.fixture
def mock_device_fails():
    """Mock device that fails to connect."""
    device = MagicMock()
    device.send_command.side_effect = Exception("Connection timeout")
    return device

Now use in tests:

def test_configure_interface(mock_device):
    """Uses fixture automatically."""
    configure_interface(mock_device, "Gi0/0/1", "10.0.0.1", "255.255.255.0")
    assert mock_device.send_command.called

def test_fails_gracefully(mock_device_fails):
    """Uses failure fixture."""
    with pytest.raises(Exception):
        configure_interface(mock_device_fails, "Gi0/0/1", "10.0.0.1", "255.255.255.0")

Running Tests & Coverage

Run All Tests

pytest tests/

Run Specific Test

pytest tests/test_decorators.py::TestRetryDecorator::test_succeeds_on_first_attempt

Run with Coverage Report

pytest --cov=src tests/

Output:

Name                Stmts   Miss  Cover
----------------------------------------
src/decorators.py      45      2    95%
src/device_ops.py      28      1    96%
src/utils.py           15      0   100%
----------------------------------------
TOTAL                 88      3    97%

Good coverage targets: 80-90% for business logic, 95%+ for critical decorators

Fail Fast While Developing

pytest -x tests/  # Stop on first failure
pytest -v tests/  # Verbose output
pytest -s tests/  # Show print statements

CI/CD Integration: GitHub Actions Example

# .github/workflows/test.yml
name: Test Automation Code

on: [push, pull_request]

jobs:
  test:
    runs-on: ubuntu-latest
    strategy:
      matrix:
        python-version: [3.9, '3.10', '3.11']

    steps:
    - uses: actions/checkout@v3
    - uses: actions/setup-python@v4
      with:
        python-version: ${{ matrix.python-version }}

    - name: Install dependencies
      run: |
        pip install -e .
        pip install pytest pytest-cov pytest-mock

    - name: Run tests
      run: pytest tests/ --cov=src

    - name: Check coverage threshold
      run: pytest tests/ --cov=src --cov-fail-under=80

Now every commit is automatically tested before merging.

Best Practices

1. Test Behaviour, Not Implementation

# ❌ BAD - Tests implementation details
def test_retry():
    @retry()
    def func():
        return "success"

    assert func.__name__ == "wrapper"  # Tests internal name

# ✅ GOOD - Tests behaviour
def test_retry():
    @retry()
    def func():
        return "success"

    assert func() == "success"  # Tests what user cares about

2. One Assertion Per Test (Usually)

# ❌ BAD - Multiple assertions make it hard to know what failed
def test_configure():
    result = configure_interface(...)
    assert result is True
    assert mock_device.send_command.called
    assert mock_device.send_command.call_count == 6

# ✅ GOOD - Separate tests for separate concerns
def test_configure_returns_true():
    result = configure_interface(...)
    assert result is True

def test_configure_sends_commands():
    configure_interface(...)
    assert mock_device.send_command.called

3. Use Meaningful Test Names

# ❌ BAD - Vague
def test_retry():
    pass

# ✅ GOOD - Describes scenario and expected outcome
def test_retry_recovers_from_transient_failure():
    pass

def test_retry_raises_after_max_attempts_exhausted():
    pass

4. Mock External Dependencies, Not Your Code

# ❌ BAD - Mocking your own code
@patch('src.device_ops.configure_interface')
def test_something(mock_configure):
    # This defeats the purpose of testing
    pass

# ✅ GOOD - Mock Netmiko (external), test your code
@patch('netmiko.ConnectHandler')
def test_configure():
    # Test your logic with real mocking
    pass

5. Parametrize Tests to Reduce Duplication

@pytest.mark.parametrize("ip,netmask,expected", [
    ("10.0.0.1", "255.255.255.0", True),
    ("192.168.1.1", "255.255.255.0", True),
    ("invalid.ip", "255.255.255.0", False),
])
def test_validate_ip_address(ip, netmask, expected):
    result = validate_ip_address(ip, netmask)
    assert result == expected

Enterprise Testing Strategy

Test Pyramid

        /\
       /  \  E2E Tests (5%)—real devices in lab
      /    \
     /      \
    /________\
   /          \
  /            \ Integration Tests (15%)—mocked device
 /              \ interactions
/________________\
/                  \
/                    \ Unit Tests (80%)—individual
/____________________\ functions in isolation

Do this:

  • 80% unit tests — Fast, cheap, comprehensive
  • 15% integration tests — Verify decorator stacks, device flows
  • 5% E2E tests (optional) — Only for critical workflows

Continuous Testing Workflow

  1. Write test before code (TDD optional but helpful)
  2. Run locally: pytest tests/ -v
  3. Push to GitHub
  4. CI/CD runs automatically
  5. Merge only if tests pass

Next Steps

Your testing foundation is now solid. Next patterns that depend on this:

Summary

Concept Why It Matters
Unit tests Catch bugs in decorators and business logic
Mocking Test without real devices or credentials
Integration tests Verify decorated functions work together
Fixtures DRY up test setup code
Parametrization Test multiple scenarios efficiently
CI/CD Catch bugs before production

Test your automation. Sleep better at night.

Need help applying this in a live Cisco environment?

This guide is part of the Nautomation Prime Foundation and stays free to read, share, and reuse. If you want the pattern implemented, governed, or adapted for your estate, that is paid engineering work — start a discovery conversation or review how Nautomation Prime delivers engagements. If you are a registered UK charity or CIC, there is a free and low-cost route instead.

Request Discovery Call View Services Charity & CIC Support