Run Playwright Tests on GitLab CI in Parallel

Bottom line: TraceLoom is a single GitLab CI job. Your pipeline still builds, lints and deploys on GitLab runners, but browser execution fans out to 50+ EC2 Spot workers in your own AWS account — with full Playwright traces stored in your S3 bucket.

Last updated: April 2026

Add to your GitLab CI pipeline

.gitlab-ci.yml

playwright:
  image: node:20
  stage: test
  script:
    - npm ci
    - npx traceloom run
  variables:
    TRACELOOM_API_KEY: $TRACELOOM_API_KEY
  rules:
    - if: $CI_PIPELINE_SOURCE == 'merge_request_event'
    - if: $CI_COMMIT_BRANCH == $CI_DEFAULT_BRANCH

Define TRACELOOM_API_KEY under Settings → CI/CD → Variables (masked, protected). No changes to your Playwright test files — TraceLoom wraps the Playwright CLI.

What you get

• ✓ 50+ parallel EC2 Spot workers. Go far beyond what a handful of GitLab runners can do in parallel, without per-session pricing.
• ✓ Full Playwright trace on every test. DOM snapshots, network waterfalls, console logs — viewable in the standard Playwright Trace Viewer.
• ✓ BYOC data ownership. Workers run inside your VPC and traces land in your S3 bucket. TraceLoom's control plane only sees run metadata.
• ✓ One GitLab CI job. Drop-in `.gitlab-ci.yml` snippet with no changes to your Playwright test files or Playwright config.

Frequently Asked Questions

How do I trigger TraceLoom from a GitLab CI pipeline?

Add a job to your `.gitlab-ci.yml` that runs `npx traceloom run`, with `TRACELOOM_API_KEY` defined as a CI/CD variable in your GitLab project settings (Settings → CI/CD → Variables, masked and protected). The full YAML is in the snippet above. TraceLoom detects non-TTY environments automatically and switches to plain-text output so GitLab CI logs stay parseable.

Does TraceLoom work with GitLab CI `parallel:` keyword?

You can, but you usually don't need to. GitLab's `parallel:` keyword spreads work across GitLab runners, which are metered. TraceLoom fans your Playwright suite out to 50+ EC2 Spot workers in your own AWS account from a single GitLab job, so you get more parallelism without provisioning more runners.

Can I use TraceLoom with self-hosted GitLab and self-hosted runners?

Yes. TraceLoom only cares that the runner can reach the TraceLoom API over HTTPS and that `TRACELOOM_API_KEY` is available as an environment variable. Self-hosted GitLab, GitLab SaaS, and self-hosted runners all work the same way. The EC2 Spot workers run in your own AWS account regardless of where GitLab itself is hosted.

Does my GitLab runner minutes bill go down with TraceLoom?

Typically yes for anything beyond a small suite. Browser execution moves off GitLab runners onto EC2 Spot instances in your AWS account. Your GitLab job still runs the trigger, any pre/post steps, and the TraceLoom CLI itself, but the heavy parallel Playwright work happens on Spot compute — usually 60–90% cheaper than the equivalent GitLab runner minutes.