Skip to main content

​
Before you start

Run your jobs on Peak labels or profiles. Change nothing else in your workflows.

​
Check what Peak keeps

Expect each job to use its own repository’s caches only.

​
Check how caches fill

Expect the first run of each job to start cold. Expect only successful jobs to save. Expect pushes to the default branch to fill the default branch cache. Expect pull requests and other branches to read their own cache first, then the default branch cache. Expect fork pull requests to start cold and save nothing. Expect actions/cache in fork pull requests to use GitHub’s cache. A cache error never fails your job. The job starts cold instead.

​
Keep Blacksmith cache steps

Keep these steps when you move from Blacksmith.
Peak serves them on its own runners. Read Workflow changes for an example.

​
Clear caches

Sign in as an Admin.
  1. Open Runners.
  2. Open Runner settings for your repository.
  3. Under Warm cache, choose Clear.
  4. Choose Clear again to confirm.
This clears the repository’s warm cache, actions/cache entries, and Docker build cache. Expect the next run of each job to start cold. Clear caches from an agent with a token that has runner write access. Use the numeric GitHub repository ID.