Skip to content

Running and uploading ​

The CLI has two ways to upload the suites of easytrace.config.json:

easytrace run -- <command>easytrace upload
Runs your testsYes, the command after --No
Which reportsOnly files written during this runEvery file that matches, whatever its age
Exit codeYour command's exit code0 (1 with --strict if the upload failed)
Use it forEveryday runs, locally and in CIUploading after a separate test step, re-uploading

Both create one build from every configured suite, or from the suites named with --suite.

The Runs page lists each uploaded suite, its result and its source.

easytrace run ​

bash
easytrace run -- pnpm test
easytrace run --suite unit -- vitest run --reporter=json --outputFile.json=.easytrace/vitest.json
  1. The CLI starts the command attached to your terminal. Ctrl+C is passed through; press it twice to stop immediately.
  2. When it ends, reports written during that run are uploaded. Older reports are skipped:
    text
    EasyTrace: no report for suite "e2e" from this run (.easytrace/playwright.json) — skipped.
  3. The CLI returns the test command's exit code. Upload problems are warnings; with --strict, a failed upload after passing tests exits 1.

To chain commands, pass a shell explicitly:

bash
easytrace run -- sh -c 'pnpm test:unit; unit=$?; pnpm test:e2e && exit $unit'

Without a config file or without credentials, the tests still run and keep their exit code; the CLI prints why nothing was uploaded.

easytrace upload ​

bash
pnpm test:json                         # tests write their reports
pnpm exec easytrace upload             # upload every suite of the config
pnpm exec easytrace upload --suite server --suite web
pnpm exec easytrace upload --dry-run   # show what would be uploaded, send nothing

upload accepts every matching file, including stale ones. Name the suites you ran with --suite, or clear .easytrace/ first.

Uploading files directly ​

With file arguments, upload ignores the config's suites and uploads those files as one run:

bash
easytrace upload "reports/*.xml" --suite integration
easytrace upload results.json --format jest-json --coverage coverage/coverage-final.json

Builds ​

Every run or upload invocation is one build:

  • In CI the build id comes from the CI (GitHub Actions run id + attempt, GitLab pipeline, CircleCI workflow, Jenkins build), so separate jobs of the same pipeline — a unit job and an e2e job — end up in the same build.
  • On a machine a new build id is generated per invocation. Uploading twice creates two builds.
  • EASYTRACE_BUILD_ID (or --build-id on upload) overrides it, e.g. to merge shards run on different machines of an unknown CI.

What the CLI prints ​

text
Uploaded 1 failed, 214 passed from 3 files (suite unit) · coverage 82.4% lines → https://app.easytrace.dev/app/runs/…
Uploaded 48 passed from 1 file (suite e2e) → https://app.easytrace.dev/app/runs/…
Build 18342917-1: 2 suites uploaded.

One Uploaded line per suite; it reports each test's final result, so a test that fails and then passes on retry counts as passed. The Build line follows only when more than one suite was uploaded. Warnings start with EasyTrace:. See Troubleshooting for each one.

EasyTrace shows Pass rate as passed final results divided by passed plus failed final results. Skipped tests are excluded. It does not show a separate first-attempt pass rate.

How EasyTrace identifies flaky tests ​

EasyTrace compares the final pass/fail result of a test with its previous completed result on the same branch. It skips skipped results and running or interrupted runs. A pass → fail or fail → pass change marks that result as a flaky change, whether it finished passed or failed. A retry inside one run does not create a flaky change; the final retry result is the run result.

For each test, EasyTrace divides flaky changes by non-skipped results on each branch in the selected period and uses the highest branch rate. It flags the test as flaky when it has at least one change and that rate meets the project's threshold (5% by default). Overview's Flaky tests rate divides flagged test cases by test cases that ran across all branches in the period. A flagged test counts once, even if it changed outcome several times.