Adding Instrumentation to Jest
Here at Sentry, we like to dogfood our product as much as possible. Sometimes, it results in unusual applications of our product and sometimes these unusual applications pay off in a meaningful way. In this blog post, we’ll examine one such case where we use the Sentry JavaScript SDK to instrument Jest (which runs our frontend test suite) and how we addressed the issues that we found.
We have high-level metrics for how well (or not) our CI is performing. This is useful to get a general sense of the health of the system, however, it does not help when doing a deeper dive into why a system is slow. We decided to instrument our CI jobs at the test runner level in order to get insights into the performance of the individual tests. We needed to be able to see if our test suite was slowing down because we were adding more tests or if it was because we had poor-performing tests.