Front50 is the system of record for all Spinnaker metadata, including: application, pipeline and service account configurations.
All metadata is durably stored and served out of an in-memory cache.
The following storage backends are supported:
- Amazon S3
- Google Cloud Storage
- SQL - recommended
SQL is a cloud agnostic storage backend that offers strong read-after-write consistency and metadata versioning.
The following types are represented in Front50 (data models):
|Application||Defines a set of commonly named resources managed by Spinnaker (metadata includes name, ownership, description, source code repository, etc.).|
|Application Permission||Defines the group memberships required to read/write any application resource.|
|Entity Tags||Provides a general purpose and cloud agnostic tagging mechanism.|
|Notification||Defines application-wide notification schemes (email, slack and sms).|
|Pipeline||Defines a reusable delivery workflow (exists within the context of a specific application).|
|Pipeline Strategy||Defines a custom deployment strategy (exists within the context of a specific application).|
|Project||Provides a (many-to-many) grouping mechanism for multiple applications.|
|Service Account||Defines a system identity (with group memberships) that can be associated with one or more pipeline triggers.|
We strive to make it easy to introduce additional metadata attributes; models are simple objects and serialized to
JSON at persistence time.
Migrators for non-trivial attribute changes are supported via implementations of the
StorageServiceSupport class maintains an in-memory cache for each metadata type and delegates read/write operations to a storage backend-specific
The following metrics are relevant to overall
| Metric | Description | Grouping |
controller.invocations (count) | Invocation counts. |
controller.invocations (average) | Invocation times. |
controller.invocations (count) | All 5xx responses. |
To start the JVM in debug mode, set the Java system property
The JVM will then listen for a debugger to be attached on port 8180. The JVM will not wait for
the debugger to be attached before starting Front50; the relevant JVM arguments can be seen and
modified as needed in
By default, Front50 is built with all storage providers included. To build only a subset of
providers, use the
./gradlew -PincludeProviders=s3,gcs clean build
You can view the list of all providers in
The tests are setup to only run if needed services are available.
S3 TCK only run if there is a s3 proxy available at 127.0.0.1:9999
This can be provided with the following command:
docker run -d -p9999:80 \ --env S3PROXY_AUTHORIZATION="none" \ --env JCLOUDS_PROVIDER="filesystem" \ --env JCLOUDS_IDENTITY="remote-identity" \ --env JCLOUDS_CREDENTIAL="remote-credential" \ andrewgaul/s3proxy
When running the S3 TCK via an IDE make sure to have env
AWS_SECRET_KEY set to
null otherwise the tests will timeout, the gradle test task is already configured this way.