CP-04 · Service
CI cache storage that sits next to your runners
CI Cache stores the caches your pipelines save and restore: dependency folders, toolchains, build directories. It keeps them in the same region as your runners, with limits sized for real projects.
- GitHub Actions
- cacheport/cache, same inputs as actions/cache
- GitLab CI
- S3-compatible runner cache
- Other CI
- cacheport cache save / restore
- Entry size
- Up to 20 GB per entry
- Retention
- Least recently used within your quota
- Scope
- Per repository and branch, with fallbacks
What you get
Everything in CI Cache
Drop-in for actions/cache
Change the uses: line. Keys, restore-keys and paths keep working.
S3 storage for GitLab runners
Point the distributed runner cache at Cacheport and every runner shares the same cache.
Self-hosted runners
Ephemeral runners start with a warm cache from the same region instead of an empty disk.
Larger limits
Quotas sized per workspace, not per repository, with entries up to 20 GB.
Branch fallbacks
Restore from the default branch when a feature branch has no cache of its own yet.
Visibility
See which keys hit, which miss and how much time restores take.
Start
Up and running in three steps
GitHub Actions
Swap actions/cache for the Cacheport action.
terminal - uses: cacheport/cache@v1 with: path: ~/.cache/pip key: pip-${{ hashFiles('requirements.txt') }} restore-keys: pip- token: ${{ secrets.CACHEPORT_TOKEN }}GitLab Runner
Configure the distributed cache in the runner's config.toml.
config [runners.cache] Type = "s3" Shared = true [runners.cache.s3] ServerAddress = "acme.cacheport.cc" BucketName = "gitlab-cache" AccessKey = "ci" SecretKey = "<CACHEPORT_TOKEN>"Any other CI
Use the CLI to save and restore any directory.
ci log $ cacheport cache restore deps-$HASH node_modules $ npm ci $ cacheport cache save deps-$HASH node_modules
| Integrations | GitHub Actions (cacheport/cache), GitLab Runner (S3 cache), CLI for Buildkite, Jenkins, CircleCI and others |
|---|---|
| Regions | Frankfurt (fra1), Ashburn (iad1), San Jose (sjc1), Singapore (sin1), Sydney (syd1) |
| Largest entry | 20 GB |
| Compression | zstd, multi-threaded |
| Retention | Least recently used within the workspace quota; entries unused for 30 days are removed |
Ecosystems
Setup guides by package manager
Common uses
Where it pays off first
Self-hosted runner fleets
Caches live next to your runners instead of in a distant cloud region.
Big toolchains
Android SDKs, CUDA libraries and compiler toolchains restore in seconds instead of downloading again.
GitLab autoscaling
Autoscaled runners share one cache instead of each starting empty.
FAQ
Questions
How is this different from Package Proxy?
Package Proxy caches individual packages from registries. CI Cache stores whole directories your pipeline saves. Many teams use both: CI Cache for the warm start, Package Proxy for whatever is missing.
Do I have to change my cache keys?
No. The GitHub action takes the same inputs as actions/cache.
Are caches shared between repositories?
No, entries are scoped per repository by default. You can opt in to shared keys for things like toolchains.
What happens when the quota is full?
The least recently used entries are removed first, so active branches keep their caches.
Works with the rest of Cacheport
Same workspace, same tokens
Start on the free plan
20 GB of cache, every registry, unlimited users. Upgrade when your pipelines outgrow it.