Troubleshoot isolated monitor failures

Problem

Your New Relic Synthetics monitor reported a failure from a single location, but the application appears to be up and running, so you're not able to identify the problem.

Solution

A single failure from Synthetics is not as limited as it may seem. It represents three consecutive failures of the same monitor. When a monitor fails, New Relic immediately queues two additional checks for the same monitor and location. Synthetics reports a failure only when all three jobs fail. (Only the third monitor run is visible in the Synthetics UI.)

The failures do not need to be the same error type. This further reinforces limitations on false positives.

Synthetics monitors may sometimes fail from a single location when the site appears to be running properly from other locations. While these failures are isolated, being able to identify issues specific to a particular location can be very valuable.

Here are some techniques to diagnose possible issues behind isolated monitor failures.

Network path

The monitor that failed may have a different network path to your site than other Synthetics locations or from locations you are testing. If there is a problem at the Synthetics location, or on the network between the location and your site, a single location may fail to return a result. This indicates that some users on similar network paths may have experienced issues.

If you see Connect timed out or Read timed out, this typically indicates that the site was unavailable or so slow to respond such that the monitor could not retrieve any data. This usually indicates an issue along the network path.

Timeouts

There may be requests that timed out or requests that are missing. You may see timeout messages such as

TimeoutError: URL {YOUR_DOMAIN} was unable to finish all network requests before reaching the maximum time limit

or

TimeoutError: Page load timed-out (unable to finish all network requests on time)

These requests could not complete before the page load timeout or the end of the script.

To solve the problem, try drilling into the resource to see when and where it responds poorly. Some resources may intermittently return slowly.

Slow page load

If the page frequently is unable to fully load within 60 seconds and you are using a simple browser monitor, try a scripted browser monitor and set a configurable page load timeout.

For more help

Write scripted browsers (created scripted browser monitors)

Recommendations for learning more: