Skip to content

Guides

How to shard tests across a matrix

One long test job dominates the pipeline.

6 min read

One long test job dominates the pipeline.

Why it happens

The suite runs serially on one machine even though the tests are independent.

How to fix it

  1. Split by historical duration rather than by file count so shards finish together
  2. Use your runner's native sharding flag where one exists (--shard, --split-by)
  3. Merge shard reports into one summary so failures stay readable
  4. Watch total cost: four shards for four minutes costs the same as one job for sixteen

Your next build could be twice as fast, at half the price

Start free. Migrating away is the same one line, and we publish that diff too.