Transport temporary data between jobs using caching
Pipelines should assume that the cache must be rebuilt
The cache keyword defines when to push and pull
One job downloads or generates data and pushes to the cache
Subsequent jobs can pull from the cache to hit the ground running
Examples for many programming languages
Runner local cache requires persistent volume for Docker executor
Cross-runner caches require an object store…
… with an S3-compatible API
Self-hosted option using MinIO
XXX https://docs.gitlab.com/ee/ci/caching/#where-the-caches-are-stored
XXX https://docs.gitlab.com/runner/executors/docker.html#configure-directories-for-the-container-build-and-cache
XXX https://docs.gitlab.com/runner/configuration/advanced-configuration.html#the-runnersdocker-section