The Test History page, which you can access under the Insights submenu, shows a visual snapshot of the results for a specific test over time. Seeing the test outcomes in a scatter plot helps reveal anomalies and patterns, which can help you identify issues with test performance and flakiness across platforms, operating systems, and browsers that you test against.
- Click the Insights tab to expand its submenu.
- Click Test History.
- Use the filters to reduce the list of tests to a limited range.
- In the Search field, enter the name of the test you want to view. If you're not sure, entering just the first few characters will bring up a list from which you can choose.
The test performance visualization displays a scatter plot of each iteration of the test that was executed during the specified date filter and matches other filter criteria. At the top of the visualization are four performance statistics for the test iterations shown:
- Total Runs - total number of test runs for the selected period
- Total Errors - total number of test runs that did not complete
- Total Failures - total number of test runs that have a recorded status of "Failed"
- Average Runtime - the mean runtime of all tests shown in the results
Below the performance statistics, the scatter plot shows each instance of the test, color-coded by run status, against the time it took the test to either execute or fail. The X-axis indicates the time range that has been selected using the time filter. The Y-axis indicates the duration of the test each time it was run. You can see the specific information about the platform, operating system, and other capabilities specified in a test by hovering your mouse cursor over the point representing it on the plot.
The charting of errors and failures in the visualization can help you get an early assessment of flaky test behavior. In this example, the test constantly fails in the first and second re-run, and succeeds in the third. This is a very typical example of a flaky behavior.
In this example, you can see how the test addOneItemtoCart was executed over the last seven days on different platforms. Along the bottom are the executions that have successfully run and passed, and have the fastest execution times. As the execution time increases, you can see that there are significantly more runs that have failed. Hovering over a few of the failed tests indicates that many of the failures are on the Chrome browser.
Applying a filter for Chrome browser, you can now see that the test is clearly failing specifically in that environment.
Comparing the two graphs, you can see that the majority of failing tests begin on November 6 and are run on Chrome, so now you can dig into what changes in your test at that time that might have contributed to these failures.