Distribute Cucumber features to workers run in parallel.
This module requires a peerDependency of the cucumber package.
npm install --save-dev cucumber cuke-skywalker./node_modules/.bin/cuke-skywalker --tags '@this and not @that'All arguments will be passed directly to the workers. Features will be pre-filtered to get an accurate feature count, which is used to chunk them for the "uniform" distribution method.
This creates a queue of features and a set of worker pipelines. These pipelines grab features one at a time until the queue is exhausted. Use this method to leverage workers more efficiently.
This chunks features into uniformly-sized groups and sends them to each worker all at once. It results in fewer output files but can leave workers with nothing to do while longer feature sets are still running elsewhere.
Options are passed as environment variables.
| Variable | Default | Description |
|---|---|---|
| CUCUMBER_PARALLEL_WORKERS | 4 | Number of worker processes to distribute features to |
| CUCUMBER_PARALLEL_REPORT_DIR | reports | Output directory for worker output JSON files |
| CUCUMBER_PARALLEL_DISTRIBUTION | roundrobin | Method for distributing features (roundrobin or uniform) |
| CUCUMBER_JS_PATH | node_modules/.bin/cucumber-js | Path to the cucumber-js bin |
cucumber-html-reporter is available to generate a decent-looking HTML reporter across all workers.
./node_modules/.bin/cuke-skywalker-reportCUCUMBER_PARALLEL_REPORT_DIR is used for the reports directory containing the worker output, and is also where
the output is written. PLATFORM and ENVIRONMENT are available for metadata values.