Menu

Integrate MSTest with Universal Agent

Overview

In this article, we will walk you through an example to integrate, execute, and schedule test executions for your MSTest test project with the Universal Agent.

Notes

The test project used in this article is stored on Github at https://github.com/QASymphony/dotnet-samples containing:

  • Dotnet core test projects that target .NET Core 2.2+: xUnitSample, MSTestSample, NUnitSample
  • Dotnet project that targets .NET 4.5 (Windows only): UnitTestSample

This article will integrate an MSTestSample project with Universal Agent.

Prerequisites

  1. Activate Automation Integration.
  2. Install and Register the Automation Host.
  3. Your machine needs to have .NET Core 2.2+ installed.
  4. This article uses Selenium that opens a Chrome browser, so make sure you also install Google Chrome to the machine where the Automation Host is running.

Step 1: Create a new Universal Agent for your MSTest Test Project

Access your Automation Host UI. Select Add to create a new agent. The New Agent dialog displays.

create-agent.png

Next, enter the following in the New Agent dialog:

General Information

  • Agent Name: enter the name of the agent, e.g. MSTest Agent
  • qTest Manager Project: select a qTest Manager project from which the agent is going to execute scheduled tests, e.g. qConnect Sample Project
  • Agent Type: select Universal Agent

Pre-Execute Script

We are going to configure a pre-execute script to either clone the sample code from github or pull the latest updates of that repo if it already exists to the host machine.

Enter the script below to the Pre-Execute Script editor, which is specific to the Operating System that the automation host is running:

Windows

if not exist "C:\dotnet-samples" (
 cd /d C:\
 git clone https://github.com/QASymphony/dotnet-samples
) else (
 cd /d "C:\dotnet-samples"
 git pull --all
)

Linux or Mac

#!/bin/bash
if [ ! -d "/usr/local/var/dotnet-samples" ]
then
cd "/usr/local/var"
git clone https://github.com/QASymphony/dotnet-samples
else
cd /usr/local/var/dotnet-samples
git pull --all
fi

Execute Command

Executor: select node as the script executor

Working Directory

  • If your host machine is running on Linux or Mac, enter /usr/local/var/dotnet-samples
  • If your host machine is running on Windows, enter C:\dotnet-samples

Execute Command

We are going to create a script written in NodeJS to:

  • obtain the automation content(s) of the scheduled test runs by resolving the value of magic variable TESTCASES_AC. Refer to this article to learn more about MAGIC VARIABLES in the Universal Agent.
  • if the value of the TESTCASES_AC variable is empty, meaning that there are no test runs being scheduled to be executed (e.g. when this Universal Agent is executed in the first time), we will let the agent execute whole tests in our MSTestSample project and then create test run logs within qTest Manager as the execution result
  • if TESTCASES_AC variable has value, being the automation contents of scheduled test runs, we will instruct the Universal Agent to only execute tests that match those automation contents

Enter the code snippet below to the Execute Command editor:

// sample Execute Command for executing .NET core test with node executor

const fs = require('fs');
const path = require('path');
const { execSync } = require('child_process');

let isWin = process.platform == "win32";

// process.env.WORKING_DIR holds the value of Working Directory configured in Universal Agent,
// if you did not specify Working Directory in Universal Agent, make sure you enter it here
let SOLUTION_DIR = process.env.WORKING_DIR || ``;
// validate the existence of SOLUTION_DIR, it it does not exist, print error and stop execution
if (!fs.existsSync(SOLUTION_DIR)) {
    console.error('No working directory found.');
    return;
}

// change .NET core version to fit your need
let TARGET_DOTNETCORE_VERSION = '2.2';

// path to test project, change it to reflect yours
let TEST_PROJECT_NAME = 'MSTestSample';
let TEST_PROJECT_DIR = path.resolve (SOLUTION_DIR, 'DotnetCore', TEST_PROJECT_NAME);
let TEST_PROJECT_PATH = path.resolve(TEST_PROJECT_DIR, `${TEST_PROJECT_NAME}.csproj`);

// possible value for configuration: Debug or Release
let CONFIGURATION = 'Debug';
// we are going to execute published test
let PUBLISH_DIR = path.resolve(TEST_PROJECT_DIR, 'bin', CONFIGURATION, `netcoreapp${TARGET_DOTNETCORE_VERSION}`, 'publish');
// path to the test project output
let TEST_PROJECT_PUBLISHED_PATH = path.resolve(PUBLISH_DIR, `${TEST_PROJECT_NAME}.dll`);

// this is the full path to dotnet command
// make sure you change it to reflect your environment
let DOTNET_EXECUTABLE_PATH = isWin ? 'C:/Program Files/dotnet/dotnet.exe' : '/usr/local/bin/dotnet';

// by default, the result folder will be created at  ${SOLUTION_DIR}/TestResults
let RESULT_DIR = path.resolve(`${SOLUTION_DIR}`, 'TestResults');
// this is the path to XML test result
let PATH_TO_XML_RESULT = path.resolve(`${RESULT_DIR}`, 'Results.xml');

// delete result dir if it exists
if (fs.existsSync(RESULT_DIR)) {
    if (isWin) {
        execSync(`rmdir /s /q "${RESULT_DIR}"`);
    } else {
        execSync(`rm -rf "${RESULT_DIR}"`);
    }
}

// the shell command builder
var commandBuilder = [];
if (!isWin) {
    // set DOTNET_CLI_HOME environment var to working directory for dotnet command to work properly
    commandBuilder.push(`export DOTNET_CLI_HOME="${SOLUTION_DIR}"`);
}

// execute `dotnet publish` command to publish the test project,
// the published results will be stored in PUBLISH_DIR
commandBuilder.push(`"${DOTNET_EXECUTABLE_PATH}" publish "${TEST_PROJECT_PATH}"`);

// copy the chromedriver to the PUBLISH_DIR dir for the Selenium test to run properly
// remove these commands if you're not running Selenium tests and so there will be no chromedriver in the output
let CHROME_DRIVER_NAME = isWin ? 'chromedriver.exe' : 'chromedriver';
let COPY_COMMAND = isWin ? 'copy' : 'cp';
let PATH_TO_CHROME_DRIVER = path.resolve(TEST_PROJECT_DIR, 'bin', CONFIGURATION, `netcoreapp${TARGET_DOTNET_CORE_VERSION}`, CHROME_DRIVER_NAME);
commandBuilder.push(`${COPY_COMMAND} "${PATH_TO_CHROME_DRIVER}" "${PUBLISH_DIR}"`);

// resolve TESTCASES_AC magic var to get the scheduled test runs
let testcases_AC = $TESTCASES_AC;
testcases_AC = testcases_AC ? testcases_AC.split(',') : [];

/**
 * Kicks off the test. What it does is to resolve the value of `testcases_AC` variable and validate:
 * Case 1: if that variable has value, meaning there is/are test run(s) being scheduled in qTest Manager
 *   -- for each scheduled test run, finds and executes the test method whose name matches that test run's automation content
 * Case 2: the value of testcases_AC is empty, meaning no test runs being scheduled when the Universal Agent is executed in the first time
 *   -- executes all the tests within the project output DLL
 */
var testMethods = '';
if (testcases_AC && testcases_AC.length > 0) {
    testMethods = '/Tests:';
    for (let testcase_AC of testcases_AC) {
        testMethods += `${testcase_AC},`;
    }
    testMethods = testMethods.slice(0, -1);// remove the last ','
}

if (testMethods != '') {
    // run specific test methods, change the log file path if desired
    commandBuilder.push(`"${DOTNET_EXECUTABLE_PATH}" vstest "${TEST_PROJECT_PUBLISHED_PATH}" ${testMethods} --logger:"nunit;LogFilePath=${PATH_TO_XML_RESULT}"`);
} else {
    // run all tests, change the log file path if desired
    commandBuilder.push(`"${DOTNET_EXECUTABLE_PATH}" vstest "${TEST_PROJECT_PUBLISHED_PATH}" --logger:"nunit;LogFilePath=${PATH_TO_XML_RESULT}"`);
}


// the dotnet test runner will throw exception when there is an assertion failed and causes the 
// test exited with code 1. So we wrap the command execution in a try catch block to ensure the 
// execute command is fully executed and not exited with code 1 when there is exception thrown from the test
try {
    // build the shell command
    let command = isWin ? commandBuilder.join(' && ') : commandBuilder.join('\n');
    // execute the shell command
    execSync(command, { stdio: "inherit" });
} catch (err) {
    console.error(`*** Test execution error ***`);
    console.error(err.message || err);
    console.error(`*** End test execution error ***`);
}

Path to Results

Enter the value below to specify the path to the test results generated when the execution completes.

  • If your host machine is running on Linux or Mac: /usr/local/var/dotnet-samples/TestResults
  • If your host machine is running on Windows: C:\dotnet-samples\TestResults

Result Parser

Select NUnit as the Result Parser. Note: if you do not find the NUnit parser in the Result Parser list, follow below steps to make it available to Universal Agent:

  • Temporarily save the MSTest Agent. You'll be return back to Automation Host dashboard
  • Download NUnit parser HERE
  • Access qTest Launch and follow this article to upload the Parser to qTest Launch
  • Go back to Automation Host UI, refresh the page for it to load the new parsers
  • Select to edit MSTest Agent and you'll find the NUnit parser from the Result Parser list

Select SAVE to finish creating the agent. Below is how MSTest Agent looks like, on Mac.

nunit-agent.png

Step 2: Execute the Agent

From Agents list, locate the MSTest Agent and click on Run Now button.

run-now.png

The agent execution dialog will display.

mstest-exec-dlg.png

Click on the Execute button to execute the agent. Once the execution completes, the Console Log shows all the test results have been submitted to qTest Manager, as highlighted in the screenshot below.

mstest-execution-result.png

Next, access to qTest Manager. Select qConnect Sample Project then go to Test Execution module. You'll see the test results submitted to qTest Manager as Test Runs under a Test Suite naming Automation YYYY-MM-DD, where YYYY-MM-DD is the date the agent is executed, as below.

nunit-execution-result.png

Now we are going to schedule test execution for some selected test runs with the MSTest Agent and verify it only executes tests that match the scheduled test runs as well submits results to those test runs.

Step 3: Schedule Test Execution for specific Test Runs

From qTest Manager, select the project qConnect Sample Project, then click on Test Execution tab and do the followings:

  1. Select the newly created test suite, in our example, it is Automation 2019-01-15.
  2. From the test run list on the right, select the first four (4) test runs
  3. Click on MORE button, then select Schedule

mstest-schedule-test.png

On the opening SCHEDULE AUTOMATION TEST EXECUTION dialog, enter the followings:

  1. Name: name of the schedule, e.g. Execute MSTest Test with Universal Agent
  2. Agent: select the MSTest Agent 
  3. Click OK button to complete test execution scheduling for the selected three test runs

mstest-schedule-test-dlg.png

Now go back to Automation Host UI. Click Poll Now button.

poll-now.png

At this stage, the Automation Host does the followings:

  • immediately polls to qTest Manager to load schedule jobs and
  • execute the job execution for the tests methods in MSTestSample project that match the automation content of the four scheduled test runs

Wait a while for the execution completes, then click on View Log icon to view execution log.

view-log-icon.png

A log screen will display that shows the latest execution logs. Verify that the log reported only two (2) tests being executed.

view-log.png

Now get back to Test Execution module of qConnect Sample Project project in qTest Manager. Click on each test run that was scheduled to be executed, you'll see the Execution History of each test run was updated with 2 test run logs: one test log created when the agent was first executed, and one test log from the scheduled execution.

mstest-scheduled-test-exec-result.png

You have successfully scheduled test execution for specific tests in your MSTest test project using Universal Agent, and have it reported the test execution to qTest Manager.

 

Powered by Zendesk