Your CI uploads large integration-test reports on every PR. Storage costs grow quickly. Which upload-artifact configuration most directly limits how long those reports consume artifact storage?
Select an answer to reveal the explanation.
Short Explanation and Infographic
retention-days on upload-artifact (and org defaults) controls artifact lifetime and cost. Step timeouts, concurrency cancel, and “artifacts never expire” are wrong models.
Full explanation below image
Full Explanation
actions/upload-artifact accepts retention-days to override default retention within platform maximums. Short retention for noisy PR diagnostics reduces storage without harming long-lived release artifacts, which can use longer retention or GitHub Releases. timeout-minutes aborts a step; it is not an artifact TTL. concurrency cancel-in-progress cancels runs, not the artifact store. Artifacts do expire based on retention policies. Pair retention tuning with path filters so docs-only PRs skip heavy report jobs, and delete artifacts via API when policy requires early cleanup.