GitHub Actions job fails with 'No space left on device' on a GitHub-hosted runner. How do I free disk space or avoid the failure?

Cause: Standard GitHub-hosted Linux runners have about 14 GB of SSD storage free for the job, and much of the image is already used by preinstalled toolchains. Large Docker images and layers, build outputs, dependency caches, test artifacts and package installs can exhaust it. Fix: 1. Find the consumer first: add a step running "df -h" and "du -sh * | sort -h | tail" (and "docker system df") at the start and just before the failing step. 2. Reduce what you create: build with multi-stage Dockerfiles, use .dockerignore, prune intermediate images with "docker system prune -af" between stages, delete build outputs you no longer need, and avoid uploading or caching huge directories. 3. Remove preinstalled software you do not use. The runner image README in the actions/runner-images repository lists the installed tools; deleting large unneeded ones (for example unused SDKs under /usr/share or /opt) in a cleanup step with sudo frees several GB. Check the README for the image you use before deleting anything and prefer well-known cleanup actions pinned to a commit SHA. 4. Use the temp area carefully: write scratch data to paths that are cleaned and keep logs small. 5. If you really need more space, use a larger runner (GitHub offers larger hosted runners with more storage) or a self-hosted runner with enough disk. 6. Self-hosted runners: clear the _work directory regularly and prune Docker images. Verify: the "df -h" step before the failing command shows enough free space and the job completes. Sources: https://docs.github.com/en/actions/reference/runners/github-hosted-runners https://github.com/actions/runner-images