Save Cache

Saves build cache using a cache key. This Step needs to be used in combination with Restore Cache.

Description

Saves build cache using a cache key. This Step needs to be used in combination with Restore Cache.

About key-based caching

Key-based caching is a concept where cache archives are saved and restored using a unique cache key. One Bitrise project can have multiple cache archives stored simultaneously, and the Restore Cache Step downloads a cache archive associated with the key provided as a Step input. The Save Cache Step is responsible for uploading the cache archive with an exact key.

Caches can become outdated across builds when something changes in the project (for example, a dependency gets upgraded to a new version). In this case, a new (unique) cache key is needed to save the new cache contents. This is possible if the cache key is dynamic and changes based on the project state (for example, a checksum of the dependency lockfile is part of the cache key). If you use the same dynamic cache key when restoring the cache, the Step will download the most relevant cache archive available.

Key-based caching is platform-agnostic and can be used to cache anything by carefully selecting the cache key and the files/folders to include in the cache.

Templates

The Step requires a string key to use when uploading a cache archive. In order to always download the most relevant cache archive for each build, the cache key input can contain template elements. The Restore cache Step evaluates the key template at runtime and the final key value can change based on the build environment or files in the repo. Similarly, the Save cache Step also uses templates to compute a unique cache key when uploading a cache archive.

The following variables are supported in the Cache key input:

• cache-key-{{ .Branch }}: Current git branch the build runs on
• cache-key-{{ .CommitHash }}: SHA-256 hash of the git commit the build runs on
• cache-key-{{ .Workflow }}: Current Bitrise workflow name (eg. primary)
• {{ .Arch }}-cache-key: Current CPU architecture (amd64 or arm64)
• {{ .OS }}-cache-key: Current operating system (linux or darwin)

Functions available in a template:

checksum: This function takes one or more file paths and computes the SHA256 checksum of the file contents. This is useful for creating unique cache keys based on files that describe content to cache.

Examples of using checksum:

• cache-key-{{ checksum "package-lock.json" }}
• cache-key-{{ checksum "**/Package.resolved" }}
• cache-key-{{ checksum "**/*.gradle*" "gradle.properties" }}

getenv: This function returns the value of an environment variable or an empty string if the variable is not defined.

Examples of getenv:

• cache-key-{{ getenv "PR" }}
• cache-key-{{ getenv "BITRISEIO_PIPELINE_ID" }}

Key matching

The most straightforward use case is when both the Save cache and Restore cache Steps use the same exact key to transfer cache between builds. Stored cache archives are scoped to the Bitrise project. Builds can restore caches saved by any previous Workflow run on any Bitrise Stack.

Unlike this Step, the Restore cache Step can define multiple keys as fallbacks when there is no match for the first cache key. See the docs of the Restore cache Step for more details.

Skip saving the cache

The Step can decide to skip saving a new cache entry to avoid unnecessary work. This happens when there is a previously restored cache in the same workflow and the new cache would have the same contents as the one restored. Make sure to use unique cache keys with a checksum, and enable the Unique cache key input for the most optimal execution.

Related steps

Restore cache

🧩 Get started

Add this step directly to your workflow in the Bitrise Workflow Editor.

You can also run this step directly with Bitrise CLI.

Examples

Check out Workflow Recipes for platform-specific examples!

Skip saving the cache in PR builds (only restore)

steps:
- restore-cache@1:
    inputs:
    - key: node-modules-{{ checksum "package-lock.json" }}

# Build steps

- save-cache@1:
    run_if: ".IsCI | and (not .IsPR)" # Condition that is false in PR builds
    inputs:
    - key: node-modules-{{ checksum "package-lock.json" }}
    - paths: node_modules

Separate caches for each OS and architecture

Cache is not guaranteed to work across different Bitrise Stacks (different OS or same OS but different CPU architecture). If a Workflow runs on different stacks, it's a good idea to include the OS and architecture in the Cache key input:

steps:
- save-cache@1:
    inputs:
    - key: '{{ .OS }}-{{ .Arch }}-npm-cache-{{ checksum "package-lock.json" }}'
    - path: node_modules

Multiple independent caches

You can add multiple instances of this Step to a Workflow:

steps:
- save-cache@1:
    title: Save NPM cache
    inputs:
    - paths: node_modules
    - key: node-modules-{{ checksum "package-lock.json" }}
- save-cache@1:
    title: Save Python cache
    inputs:
    - paths: venv/
    - key: pip-packages-{{ checksum "requirements.txt" }}

⚙️ Configuration

Inputs

Key	Description	Flags	Default
key	Key used for saving a cache archive. The key supports template elements for creating dynamic cache keys. These dynamic keys change the final key value based on the build environment or files in the repo in order to create new cache archives. See the Step description for more details and examples. The maximum length of a key is 512 characters (longer keys get truncated). Commas (,) are not allowed in keys.	required
paths	List of files and folders to include in the cache. Add one path per line. Each path can contain wildcards (* and **) that are evaluated at runtime.	required
verbose	Enable logging additional information for troubleshooting	required	false
compression_level	Zstd compression level to control speed / archive size. Set to 1 for fastest option. Valid values are between 1 and 19. Defaults to 3.		3
custom_tar_args	Additional arguments to pass to the tar command when creating the cache archive. The arguments are passed directly to the tar command. Use this input to customize the behavior of the tar command when creating the cache archive (these are appended to the default arguments used by the step). Example: --format posix
is_key_unique	Enabling this allows the Step to skip creating a new cache archive when the workflow previously restored the cache with the same key. This requires the cache key to be unique, so that the key changes whenever the files in the cache change. In practice, this means adding a checksum part to the key template with a file that describes the cache content (such as a lockfile). Example of a cache key where this can be safely turned on: npm-cache-{{ checksum "package-lock.json" }}. On the other hand, my-cache-{{ .OS }}-{{ .Arch }} is not unique (even though it uses templates). Note: the Step can still skip uploading a cache when this input is false, it just needs to create the archive first to compute its checksum (which takes time).		false

Outputs

There are no outputs defined in this step

🙋 Contributing

We welcome pull requests and issues against this repository.

For pull requests, work on your changes in a forked repository and use the Bitrise CLI to run step tests locally.

Note: this step's end-to-end tests (defined in e2e/bitrise.yml) are working with secrets which are intentionally not stored in this repo. External contributors won't be able to run those tests. Don't worry, if you open a PR with your contribution, we will help with running tests and make sure that they pass.

Learn more about developing steps:

• Create your own step
• Testing your Step