Member-only story
Optimizing Jest for Faster CI Performance with GitHub Actions
Running tests for a front-end application is no trivial task. There’s TypeScript to compile, dependencies to mock, a virtual browser to run, and code coverage to track. Jest is a powerful ecosystem, but all of this functionality comes with a cost. Jest can be very slow, and while its default configuration and behaviors help optimize it when running in your local development environments, in your Continuous Integration (CI) pipelines, it’s another story. Without attention, Jest is a significant contributor to poor CI pipeline performance.
Fortunately, a couple of simple techniques will dramatically improve Jest’s performance in your CI pipelines, letting you reduce your overtime pipeline runtime down to whatever number you want it to be.
But before we get to that, let’s cover some of the impacts slow CI performance might be causing your team. Too often, I have seen teams content with slow pipelines, unaware of the impacts that it is having on their team.
Why improving Continuous Integration performance is worthwhile
1) Delayed feedback interrupts other tasks
Our CI pipelines provide essential feedback on our code — running static analysis, test automation, and coverage. Ideally, we get this feedback before we get too deep into another task. Feedback that comes after we have already switched to another task means choosing between context shifting out of that task or delaying fixing the highlighted issues.
2) Multiple people merging at the same time get blocked
If you’re using Branch Protection Rules and GitHub Required Statuses, your CI pipeline needs to finish before you can merge with the main branch. The longer your CI takes to run, the greater the chance of someone beating you to the merge, forcing you to fetch the latest changes and restart your merge. This situation is even more prevalent for teams using Monorepos, where multiple teams are committing to one repository, or teams using Trunk-Based Development, where each team member might merge with the main branch several times a day.
GitHub is working on a merge queue that will alleviate these problems, but in the meantime, we need to keep an eye on…
