Getting Started with Transactions

Introduction

ThousandEyes transaction tests are Web layer tests similar to HTTP server tests and page load tests, except that transaction tests can interact with their targets in order to mimic multi-step user journeys or API sequences. Transaction tests provide greater insight into user experience, and enable you to ensure that those user journeys complete successfully.

Compare these three web layer test types below:

HTTP Server Test	Page Load Test	Transaction Test
The ThousandEyes Cloud or Enterprise Agent sends a single request and expects a single valid response. Measures response time and validates the HTTP status code to compute server availability. HTTP server tests are used to monitor web servers to ensure they are available and performant.	ThousandEyes agent runs Chromium in order to request and render a single web page. Chromium sends an initial request for the page, then renders the response, and iteratively requests all of the components within the page, including images, JavaScript files, CSS, and AJAX requests. Page load tests monitor the web server that serves the page and all of the page's dependencies. A page load test can be thought of as multiple HTTP server tests combined in one.	Browser Synthetics: Agent runs Chromium which is automatically driven by a Selenium script. The script can navigate through multiple pages and interact with them. A browser synthetics transaction test can be thought of as multiple page load tests combined in one. API Monitoring: Agent runs a script in Node.js to sequentially or iteratively make arbitrary API requests towards the target using the node-fetch library.

When to Use a Transaction Test

Transaction tests measure web user experience using either synthetic browser interactions, or sequences of API requests. Transaction tests should be used for testing multi-step workflows. This type of test can uncover problems that aren't always apparent from loading a single page, as with a page load test, or sending a single request, as with an HTTP server test. For example, if your app or web site relies on returning customer data from somewhere else after the user logs in, you'll need a transaction test to evaluate this part of the user experience past the login screen.

Some examples of transaction tests include:

• Productivity SaaS: Log in, browse to a shared documents folder, and download a file.

• Shopping/e-commerce: Load the main page, search for a specific product by name, add it to a shopping cart, and complete the purchase using a dummy credit card.

• Web conferencing: Log in, schedule a meeting, and join the meeting in the browser with a virtual camera and microphone

Terminology

• BrowserBot is a component of the ThousandEyes Cloud Agents and Enterprise Agents that manages page load and transaction tests. This is accomplished by running an instance of the Chromium web browser which can be driven automatically via Selenium from Node.js. Complete details are available in the What is BrowserBot? article.

• Selenium is an open source software library for browser automation. Transaction tests leverage the Selenium API in Node.js for automating the Chromium web browser.

• node-fetch is an open source software library for making HTTP requests. Transaction tests leverage node-fetch within Node.js for API monitoring, allowing you to script machine-to-machine workflows such as back-end API call sequences.

• Node.js is a JavaScript runtime environment included in BrowserBot which is used to execute scripts for transaction tests.

• JavaScript is the programming language in which transaction scripts are written. You do not need to be a JavaScript expert to work with transaction tests, but some JavaScript knowledge will allow you to create more advanced transaction scripts.

Getting Started with the Recorder IDE

Requirements

The Recorder IDE has two modes, based on your permission settings. Both modes require the API Access permission and either the Login via ThousandEyes login page or Login via Single Sign-On permission.

• Standalone Mode: Allows the user to use the IDE for standalone test creation. No additional permissions required.

• Test Creation Mode: Allows the user to create tests and export (upload) the tests to the ThousandEyes platform at app.thousandeyes.com. This mode additionally requires the Edit tests, Create web transaction tests, View agents in account group, and View labels permissions.

Refer to the ThousandEyes Recorder Permissions article for full details.

Downloading and Installing the Recorder IDE

• Windows - Windows 7 or later - Download - Windows exe
• macOS - MacOS 10.10 (Yosemite) or later - Download - Mac dmg
• Linux - Ubuntu 12.04 or later - Download Linux AppImage

Follow the normal installation process for your computer's operating system. For example, the macOS version of the ThousandEyes Recorder downloads as a disk image (.dmg) file. Double-click the downloaded file and a macOS installation dialog opens. In that dialog, drag the application icon into your Applications folder.

First Time Setup

After completing the installation, launch the ThousandEyes Recorder IDE. The first time you use the Recorder IDE, you will be prompted to login to your ThousandEyes account. Type your username and password and click the Log In button, or click Single sign-on if your organization is configured to use an external identity provider for authentication.

ThousandEyes login screen

After completing the login flow, the Recorder IDE may flash briefly while it verifies your user has the necessary permissions. After this, it should look like the screenshot below.

The ThousandEyes Recorder IDE

If the Recorder IDE shows the message "You Do Not Have Permissions to Create Tests", as seen below, double check that your ThousandEyes user is assigned a role with the requisite permissions.

Recorder IDE screen, displaying a restricted permissions warning

You may also change your current account group by clicking on your e-mail address in the top right corner.

Dropdown menu for changing the current account group

For complete details on using the ThousandEyes Recorder IDE, refer to the ThousandEyes Recorder article.

Browser Synthetics Development Workflow

Flowchart of Browser Synthetics development

Recording a Browser Transaction

Click the Record button () and the Recorder IDE will prompt you to type a URL to record. You may also configure the browser window size and orientation. After typing a URL, click Start Recording and the Recorder IDE will launch a Chromium browser window which will automatically navigate to that URL.

A screen recording example using the Google website

The browser will record your interactions, such as clicking on buttons or typing text in form fields, and use them to automatically generate a transaction script. When you have finished your recording, go back to the Recorder IDE window and click the Stop button ().

As an example, click the Record button and type https://google.com in the Enter Base URL field, then click the Start Recording button. When the Chromium window has opened and loaded https://google.com, click on the search field, then type "ThousandEyes product documentation", and then press the "Enter" key. On the search results page, click the link to the ThousandEyes product documentation site. Then go back to the Recorder IDE window and click the Stop button.

Example of generating a transaction script with the recorder

After clicking the Stop button, the Recorder IDE will close the Chromium window and generate a transaction script for automating the workflow you just recorded. In this example, you should expect to see a script generated like the one below:

import { By, Key } from 'selenium-webdriver';
import { driver, test } from 'thousandeyes';

runScript();

async function runScript() {
  
   await configureDriver();

   const settings = test.getSettings();

   // Load page
   await driver.get(settings.url);
  
   // Click on 'Search'
   await click(By.name(`q`));

   await typeText('thousandeyes product documentation', By.name(`q`));

   await pressEnter(By.name(`q`));

   // Click on 'ThousandEyes Documentation - Thousa...'
    await click(By.css(`[href="https://docs.thousandeyes.com/"] > .LC20lb`));
  
}

You have now successfully recorded your first transaction test. The next step in the browser synthetics development workflow is to play the test to verify it works as expected. If you would like a better understanding of the generated code for this transaction script, continue on to the next section below. Otherwise, jump ahead to the Playback section.

Breaking down the script line by line

The first two lines of the script are import declarations. The transaction script runs in Node.js, a JavaScript runtime environment, and allows us to import code from outside of the script. In this case, the Recorder IDE automatically imported code from two packages named "selenium-webdriver" and "thousandeyes".

import { By, Key } from 'selenium-webdriver';

The first declaration, shown above, imports a class named By and an enumeration named Key from Selenium. The By class is used for specifying the method for locating an element within a webpage. To interact with an element, like clicking on it or typing into it, Selenium must first be able to locate the element. Elements will be generally located in one of the following ways:

• className: Locates elements that have a specific class name.

• css: Locates elements using a CSS selector

• id: Locates elements by the ID attribute

• linkText: Locates link elements whose visible text matches the given string

• name: Locates elements whose name attribute has the given value

• xpath: Locates elements matching a XPath selector

The Key enumeration provides representations of pressable keys that aren't text, such as the Alt, Shift, Tab, and Enter keys. The Recorder IDE imported this enumeration because we pressed the Enter key while recording.

import { driver, test } from 'thousandeyes';

The second declaration imports two modules provided by ThousandEyes. The first module, driver , is an instance of Selenium's WebDriver class that has been instantiated to work with the Chromium browser on the agent. Refer to the Controlling the Browser section of the Transaction Scripting Reference for the full list of supported methods. The second module, test, provides an interface to the transaction test configuration, such as getting the target test URL or test interval. Refer to the Getting Test Configuration Settings section of the Transaction Scripting Reference for full documentation and example usage.

runScript();

The line above calls the runScript function, which contains the interactions that were recorded. The runScriptfunction is defined in the following lines.

async function runScript() {

This line begins the definition for runScript function.

   await configureDriver();

This line calls the configureDriver function which is provided by the Recorder IDE. The configureDriver function configures the driver instance's implicit timeout. The implicit timeout specifies the maximum amount of time to wait when attempting to locate elements on the page. Without calling configureDriver, the default value is 0 which is not very forgiving on modern websites and will often throw errors that elements could not be found because the page had not fully loaded. When configureDriver is used, the value is set to 7 seconds.

  const settings = test.getSettings();

This line calls the getSettings method from the test module and stores the result in a variable named settings.

  await driver.get(settings.url);

This line calls the get method of the driver instance to navigate to a specific URL. It references the settings variable from the previous line to use the target URL from the test configuration.

  await click(By.name(`q`));

This line is generated from clicking on the search field on the Google search page. It calls a function named click, which is provided by the Recorder IDE, with an element selector using By.name, which was imported earlier from Selenium. The value of the name selector, "q", corresponds to the HTML name attribute of the search field on Google’s page.

  await typeText('thousandeyes product documentation', By.name(`q`));

The typeText function is provided by the Recorder IDE as a helper function for typing text into a given field. It accepts two parameters: first, the text to type, and second, a selector to locate the element in which to type that text.

  await pressEnter(By.name(`q`));

The pressEnter function is provided by the Recorder IDE as a helper function for pressing the enter key. It accepts one parameter, a selector to locate the element in which to press the Enter key.

  await click(By.css(`[href="https://docs.thousandeyes.com/"] > .LC20lb`));

The last line in the runScript function is generated from clicking on the ThousandEyes Product Documentation link on the search results page. Here, the Recorder IDE used a different selector, By.css to locate the link.

}

The closing curly brace concludes the definition of the runScript function.

The helper functions mentioned above that are generated and included by the Recorder IDE, such as configureDriver, click, typeText, and pressEnter, are defined below the the definition of the runScript function.

Playback

Once you have recorded a transaction, you can play the script to test that it works as expected. Click the Play button () and the Recorder IDE will launch Chromium to execute the Transaction script.

Playing back a generated script to test it

If everything worked correctly, the Chromium window will close and no errors will be reported. If your test playback worked successfully, then you should optimize the script by adding markers, screenshots, and configuring credentials, then export the script as a scheduled transaction test in the ThousandEyes platform.

If any errors were encountered while executing the script, they will be reported in the Recorder IDE’s console located in the bottom pane of the window, as seen below. When an error occurs, the Recorder IDE will also capture a screenshot from the browser and highlight the lines of code leading to the error. If your test playback encountered errors, then you should proceed to troubleshooting the script so that it correctly and completely emulates the user journey without errors.

An example of errors displayed in the Recorder IDE's console

Optimize

Once your transaction script successfully plays back without any errors, you should optimize the script with the features described below. After adding your optimizations, be sure to play the transaction again to verify it still works as expected.

Markers

You can use markers to define and measure discrete steps within a user journey or business transaction. For example, an e-commerce checkout transaction may include steps for "Item Search", "Add to Cart", and "Submit Order", and each step may include multiple actions like clicking and typing. Markers are used to delineate such steps and measure the time taken for each. By identifying where a script section starts and stops, markers can be compared to over test rounds and to the overall transaction time. If the performance of the whole transaction degrades, markers allow you to quickly see which step(s) in the transaction took longer to complete than they normally would. When an error occurs inside of a marker, the marker will show as "Incomplete" in the test view which can identify the specific phase of the transaction that had an error. Marker times are displayed on the timeline and waterfall in transaction test views and can also be used in alert rules and dashboards.

Diagram displaying Search and Add to Cart transaction steps and the length of time for each step

There are two ways to use markers:

• the set method creates a marker with the supplied name that spans from transaction start time to the time this method is called

• the start and stop methods start and stop a marker with the supplied name at the time they are called

To use markers in your script, you must include the markers module in the thousandeyes import line at the top of the script.

import { driver, markers, test } from 'thousandeyes';

Clicking the Add marker button () will:

• Add markers to the import { ... } from 'thousandeyes' import declaration, if the module is not already imported

• Add two lines of code for markers.start and markers.stop at the cursor position in the script editor

The markers module is documented in the Transaction Scripting Reference. The code below shows how to add markers to the example script from the Recording a Browser Transaction section above.

import { By, Key } from 'selenium-webdriver';
import { driver, markers, test } from 'thousandeyes';

runScript();

async function runScript() {
  
  await configureDriver();
  const settings = test.getSettings();
  await driver.get(settings.url);  

  // Using markers.set, the ‘Initial Page Load’ marker spans from transaction start time 
  // to the time the initial page finishes loading
  markers.set('Initial Page Load');
  
  // Using markers.start, the ‘Search’ marker begins measuring time after the initial page loaded
  markers.start('Search');
  await click(By.name(`q`));
  await typeText('thousandeyes product documentation', By.name(`q`));
  await pressEnter(By.name(`q`));
  // Using markers.stop, the `Search` marker ends when the user submits the search form
  markers.stop('Search');  


  markers.start('Product Docs');
  await click(By.css(`[href="https://docs.thousandeyes.com/"] > .LC20lb`));
  markers.stop('Product Docs');
}

Screenshots

Screenshots are useful to track progress and ensure the page visually matches what you were intending to see. You can capture a screenshot of the browser’s viewport by adding the following line of code inside your script. Clicking the Take screenshot button () will add this line of code at the cursor position in the script editor. No additional import declarations are required because the takeScreenshot method belongs to the driver class that has already been imported after the recording.

  await driver.takeScreenshot();

There is no limit to the number of screenshots that can be captured during a transaction, but only the last three screenshots will be included with the test results in the platform as described in Screenshots in Transaction Test Views. It can still be useful to capture more than three screenshots because if the script encounters an error before completing, you can inspect the state of the webpage leading up to the error. One screenshot will be automatically captured at the time the error occurs.

Capturing screenshots requires a small-but-non-zero amount of time on the order of one half second. Whenever possible screenshots should not be captured gratuitously and should be captured after stopping one marker and before starting the next.

You can disable screenshots by unchecking the Screenshots > Enabled checkbox in the test's Advanced Settings tab in the ThousandEyes platform. Unchecking this box will prevent screenshots from being captured when driver.takeScreenshot() is called or when an error occurs.

Credentials

The Recorder IDE’s Chromium browser will record the keys you press while interacting with a page. Recall from the example Google search transaction that typing in the search form resulted in this line in the transaction script:

  await typeText('thousandeyes product documentation', By.name(`q`));

In some cases, such as typing in password forms, you may not want to store the values you typed as cleartext in the transaction script. ThousandEyes provides a credential repository which is used to store and retrieve sensitive information like passwords, authentication tokens, or two-factor authentication secrets.

The Recorder IDE will automatically use the credentials repository if you record typing in an input field with a type attribute equal to "password". As an example, start a new recording on the URL https://app.thousandeyes.com and when the Chromium window opens and loads the ThousandEyes login page, click on the Password input field and type some text, and then click the Log In button. Switch back to the Recorder IDE window, click the Stop button.

Notice how the call to the typeText function differs from the earlier example:

  await typeText(credentials.get('pass_1674485288249'), By.id(`password`));

The Recorder IDE detected we were typing in a "password" field and automatically stored the value as a local credential rather than using a plaintext string. Instead of seeing the characters you typed, the first parameter of the typeText function is now a call to credentials.get with a credential name of "pass_1674485288249". The credential name is generated using the current unix timestamp to avoid naming collisions with other credentials. While you can use the credential with its default name, the recommended best practice is to update the credential name to something meaningful or identifiable.

Credentials can be managed in the Recorder IDE’s credentials dialog by clicking the Show credentials button (). This dialog contains two sections: the credential repository from the ThousandEyes platform and local credentials (those that have not yet been synced to the ThousandEyes platform). To rename a credential, hover over the credential name in the list, then click the pencil icon. Then, you can edit both the name and the value of the credential. Click the Update Credential button to save your changes.

Credentials List	Credential Details

After you rename a credential, be sure to modify the name in any calls to credentials.get that used the old name. For example, if you rename the "pass_1674485288249" credential to "myPassword", then you must update the typeText line from above like so:

  await typeText(credentials.get('myPassword'), By.id(`password`));

You can also sync the credential to the ThousandEyes platform by hovering over the credential name in the list and clicking the Save to ThousandEyes icon. If you plan to export this transaction to the platform, you should sync your credential now.

Troubleshooting

You may find that some recorded transactions will raise an unexpected error during playback in the Recorder IDE or while running on Cloud and Enterprise Agents in the ThousandEyes platform. While many recorded flows will result in fully functional scripts that work on the first playback, at times it will be necessary to manually alter the recorded output. This section will help you to troubleshoot your transaction scripts to identify the cause of errors and how to fix them.

To most effectively troubleshoot transaction scripts, you should have some familiarity with the browser's web development tools. See the Working With Web Development Tools article for more information and resources.

Useful Functions for Troubleshooting

console.log

You can use console.log statements to help debug, such as printing the current line number, e.g. console.log("Line 7"), or the value of a variable, e.g, console.log(settings.url).

driver.sleep

You can use driver.sleep to slow down script execution when debugging or if you want to sit at one step for a long time (for instance, to inspect elements in the page using the web development tools). The driver.sleep function takes a single parameter, the amount of time to sleep in milliseconds. Example usage:

    // Sleep for 5 seconds
    await driver.sleep(5 * 1000);

Clicking the Add sleep button will insert code to call driver.sleep at the cursor position in the script editor.

Common Errors

NoSuchElementError

The most common error is NoSuchElementError which indicates that Selenium was unable to locate an element within the page with the given selector. This usually occurs for one of the reasons explained below.

Element did not exist (yet)

Today's web applications commonly use client-side rendering to build their pages. Compared to server-side rendering, which generates the full HTML document and sends it to the client in the initial request, with client-side rendering the server only sends an initial scaffold in the first request and then uses JavaScript to dynamically fetch data and render it in the page. Selenium will begin its attempt to locate an element after the initial page load, but does not necessarily wait for all the JavaScript to finish executing.

If the transaction is failing because it cannot locate an element, you can try increasing the implicit value in the configureDriver function. For example, the code below sets the implicit wait to 15 seconds, a little more than double the Recorder IDE's default value. Note that the value is in milliseconds, which is why * 1000 is used.

async function configureDriver() {
    await driver.manage().setTimeouts({
        implicit: 15 * 1000, // If an element is not found, reattempt for this many milliseconds
    });
}

Element did exist, but selector did not match

Some pages contain elements that do not have consistent attributes which may cause the Recorder IDE to choose a selector that matches while recording but does not match during playback. There are two easy ways to identify if this is the cause of your NoSuchElementError:

1. Use the console.log and driver.sleep functions immediately before the line of code which raises the NoSuchElementError and play the transaction again. After the browser opens, switch back to the Recorder IDE and when you see your log message in the Recorder IDE's console, hit the Pause button. Return to the browser and use the web development tools to inspect the element, comparing its attributes with the selector used in the script. It may be helpful to repeat this more than once.

2. Re-record your transaction and compare the selector(s) between the first and second recordings. Make sure to copy or save the original recording before starting a new one because the new recording will overwrite the original script in the script editor.

Element was in a different tab or window

Some browser actions may cause a new browser tab (or window) to open. While the Recorder IDE will continue to record your actions in other tabs, it does not record when you switch between them. This can lead to the script looking for an element in one page when it exists in another page in a different tab.

To change the active tab, you can use the driver.switchTo function. Two examples are available in the transaction scripting examples repository:

1. switchToNextTab.js: Shows how to switch to the next tab, including wrapping around to the first tab from the last tab
2. switchToTabWithUrl.js: Shows how to switch to a specific tab given its URL

ElementClickInterceptedError

The ElementClickInterceptedError will occur when Selenium attempts to click on an element but it is covered by one or more other elements. This can happen when the web page uses the CSS z-index property to overlay elements on top of other elements. This normally occurs when a popup dialog is shown or when sticky navigation bars or footers scroll with the page.

To avoid this error, be sure to wait for and dismiss any automatic modal popups while recording. If the target element is covered by a sticky navbar or footer, try using the scrollElementIntoView function from the from the transaction scripting examples repository.

ElementNotInteractableError and WebDriverError: element not interactable

These errors will occur when Selenium attempts to interact (e.g, click on or type into) with an element that is not visible. In contrast to the NoSuchElementError, the element does exist and was successfully located, but Selenium cannot interact with it because it is not rendered in the page.

This commonly occurs when interacting with an element that is only visible while hovering the mouse pointer over a particular part of the page. The Recorder IDE does not record mouse hover events, so you will need to add the code to move the mouse pointer. There are two ways you may be able to resolve this issue:

1. Re-record the transaction and try clicking the element, not just hovering over it. Many pages will handle clicks and hovers the the same way, though some may handle them differently.

2. Use the moveMouseInto function from moveMouseIntoElement.js in the transaction scripting examples repository to move the mouse pointer to a specific element and trigger hover events.

TimeoutError: Transaction timed out

This error occurs when the configured test timeout is reached before the script has completed. To prevent this error, you can increase the test timeout or reduce the runtime of the script. For example, you can make the script run faster by minimizing or removing use of driver.sleep. If your transaction is lengthy or complex with lots of steps, consider breaking it down into two or more separate transactions.

The maximum configurable timeout is dependent on the test interval:

• 60 seconds for 2 minute tests

• 150 seconds for 5 minute tests

• 180 seconds for 10+ minute tests

To configure the timeout in the Recorder IDE, click the Show settings button () and adjust the Timeout slider in the Advanced Settings tab. To configure the timeout in the ThousandEyes platform, go to the Test Settings page, find and expand the test in the list, and adjust the Timeout slider in Transaction Timing section of the Advanced Settings tab.

Export to ThousandEyes Platform

After you've recorded, optimized, and played back your transaction, you are ready to export it from the Recorder IDE to the ThousandEyes platform. Click the Export to ThousandEyes button to open the Test Settings dialog where you can select the test interval and agents, associate alert rules, and configure other advanced settings.

Basic Settings	Browser Options	Advanced Settings

As mentioned in the Requirements section above, exporting to the ThousandEyes platform requires the necessary permissions to use the Recorder IDE in Test Creation Mode. If you are using the Recorder IDE in Standalone Mode, you can save your transaction script locally to share it with your ThousandEyes administrator. Click the More actions button , then click the Save Script menu item.

When you are finished configuring the settings, click the Export button and the Recorder IDE will create a new test in the ThousandEyes platform. When it's finished, you should see a confirmation dialog indicating the test was successfully saved. Click the View test link in this dialog to open the test settings in platform and validate the test.

Dialog confirming test creation was successfully saved

If you are using the Recorder IDE in Test Creation Mode and your script uses any credentials which have not been synced to the ThousandEyes platform, you will receive a confirmation dialog to create the test. You can either click the Cancel button and sync the credentials before creating the test, or you can click Create Test and then sync the credentials afterwards.

The Test Settings - Credentials dialog

Validate

It is important to validate that the transaction runs successfully on the agents in the platform even though you have already played back the transaction locally in the Recorder IDE. The Recorder IDE is very closely related to the BrowserBot that runs inside the ThousandEyes Cloud or Enterprise Agent, but the they are not identical. This means that in some transactions, a script which successfully played back in the Recorder IDE may fail when run from agents in the platform.

After you export your test, click the View test link in the success dialog to open the test settings in platform. At the bottom of the test settings panel, click the Run Once button to run an instant test in a new browser tab. Wait for the test results and then verify that all of the agents successfully completed the transaction. If any errors occur, click the Run Again link above the timeline in the view to determine if the error is consistent or intermittent. Then, refer back to the troubleshooting section to determine the cause of the error.

My script works in the recorder, but fails in the platform. Why?

User Agent: The default curl user agent we use for HTTP tests is rejected by many larger sites. You should supply a chrome user agent or similar so that all layers use the same UA. You can configure the user agent in the HTTP Request section of the Advanced Settings tab in the test's settings.

Agent Location: Scripts that work fine locally may fail on agents in other regions, as they may be routed differently, land on different sites etc. Make sure you rule out any location specific issues.

Agent Network Conditions: The time that network requests take can have an impact on whether your script works or not. For example, if you have a driver.sleep command that waits for 5 seconds, that may work for you but not be sufficient for an agent in some far off rural area. It's best to not use hard coded sleep values, but instead use waits with sufficient timeouts.

Authentication: If you test an application that uses Integrated Windows Authentication (IWA), it may behave differently while recording in Windows than it does while running on the Linux-based cloud and enterprise agents. IWA, sometimes called "silent authentication", uses your Windows credentials to automatically authenticate to a web server without an interactive login form, resulting in a different transaction workflow from a Windows device than a cloud or enterprise agent. IWA is commonly used for internal applications or intranet sites, but may also be used with SaaS applications when they are configured with federated single sign-on.

Network Security: When the transaction runs on the agent, it is almost always from a different network than where it was originally recorded. This can cause different firewall or proxy rules to apply and block the traffic coming from the agent. If the transaction test completes during playback in the Recorder IDE but fails in the ThousandEyes platform, you should check the network layer metrics and path visualization to ensure the agent is able to reach the target. For Enterprise Agents, you may need to configure a proxy or verify that the necessary fire rules are in place.

Using the Transaction Test View

When you export a Transaction test, you can see the test data under Network & App Synthetics > Views. The Transaction view leverages the ThousandEyes standard layout, documented here. For general information on the ThousandEyes standard view layout, see Getting Started with Views. This section highlights specifics shown in the Transaction view.

Transaction View, Timeline

The Transactions view, showing the timeline

The top of the Transaction view shows a timeline similar to other views, with a time period slider below. Three metrics can be displayed on the timeline:

• Transaction Time (default): The amount of time spent running the transaction test. When Transaction Time is selected, you may also select up to three markers to view on the timeline

• Errors: The number of timeout errors, page errors, assert errors, and other errors that occurred in the test round

• Completion: Whether or not the transaction completed without errors

When a single agent is selected, you can mouse over the error indicator in the swimlane under the timeline to view the specific error details.

Transaction View, Map

The Transactions view's map tab, showing global results from agents running tests

The Map tab as shown below displays a global map showing all of the agents that are running the test, with a Metrics panel to summarize data and any error messages reported by the agents. Clicking on an agent in the map will select that agent in the Agent selector above the timeline. The color of the agents shown on the map indicates the speed relative to the test duration, as specified by the Timeout parameter in the Advanced Settings tab for the test configuration in Network & App Synthetics > Test Settings.

The Metrics panel shows the transaction time (average, when all agents are selected) and when a single agent is selected also shows the individual marker times.

The Metrics panel, showing global agents' transaction times

Transaction View, Table

The Transaction view's Table tab, showing test results from selected test rounds

The Table tab shows test details from the selected test round for each agent. If any errors occur during the transaction, the agent will be shown in the table with a red dot next to its name. You can mouse over this icon to view specific error details. Clicking on an agent in the table will select that agent in the Agent selector above the timeline.

For more details on the transaction view table, see the Transaction Test Table Tab View article.

Transaction View, Waterfall

The Transaction view's Waterfall tab, showing details about pages, markers, and screenshots

The Waterfall tab in the transaction view is similar to the Waterfall tab in the page load view, with three additional features at the top of the chart:

• Pages: Because transaction tests can navigate through more than one page, the waterfall chart in the transactions view indicates each page that is visited and the duration of time spent on that page. You can mouse over a page in the Page row to view the page's duration. You can click on a page to filter the waterfall chart to only include components from that page.

• Markers: Markers are exclusive to transaction tests. The waterfall chart for transactions includes a visualization of the marker timings over the duration of the transaction. You can mouse over a marker in the Markers row to view the marker's exact timing. You can click a marker to filter the waterfall chart to only include components during that marker.

• Screenshots: The last three screenshots captured during the transaction will be stored with the test data and displayed on the Waterfall tab. You can mouse over the picture icon in the Screenshots row to view the captured screenshot. The horizontal position of the icon in the row indicates the relative point in time during the transaction that the screenshot was captured.

For complete details on the Waterfall tab, see the Navigating Waterfall Charts for Page Load and Transaction Tests article.

Resources

ThousandEyes Product Documentation

• What is BrowserBot?
• ThousandEyes Recorder
• Role-Based Access Control, Explained
• Transaction Scripting Reference
• Working With Secure Credentials
• Working With Web Development Tools
• ThousandEyes Standard Layout
• Getting Started with Views
• Transaction Test Table Tab View
• Navigating Waterfall Charts for Page Load and Transaction Tests

External Resources

• Troubleshooting synthetics scripts
• Selenium javascript API documentation