-
Notifications
You must be signed in to change notification settings - Fork 20
high level overview
This document is designed to provide a high level understanding for different components of the operator. Content covers: what specific controllers do, why specific controllers are important, and basic architecture of the project (how different components interact with each other).
| Folder | Purpose |
|---|---|
| assets | Stores static assets used in the operator |
| build | Build output. |
| bundle | Temporary directory to use for building the CSV bundle. |
| cmd | Commands to run the operator |
| deploy | Deploy specific code included CSV and generated CRDs |
| docs | Documentation |
| /Makefile | Make commands to simplify development |
| pkg | The business logic for the operator |
| pkg/apis | API files to generate CRDs for the operator |
| pkg/controller | Controller codes |
| pkg/managers | Code to bootstrap new controller managers |
| pkg/reporter | Montioring reporter specific code |
| pkg/utils | Helpful utilitly functions |
| reports | Temporary directory for reporting |
| scripts | Contains scripts to help build or use the operator |
| test | All our integration or end-to-end tests. Also includes testing tools |
| test/e2e | End to end testing |
| test/rectest | Contains custom and generated code to simply unit testing |
| version | Version file contains the operator version |
MarketplaceConfig is the first CR created. Its prime responsibility is to deploy the remaining CRs. MarketplaceConfig decides which CRs to create based on flags. Currently MarketplaceConfig can create:
- RazeeDeployment
- MeterBase
- The Operator Source
- The IBM Catalog Source
MarketplaceConfig is valuable because it is a single point of origin that ensures the correct resources are installed on the cluster.
MeterBase is responsible for setting up prometheus. MeterBase currently creates:
- The Prometheus Operator
- Service for Prometheus
- Persistant Volume Storage
MeterBase is valuable because it ensures we can track metrics via prometheus.
WIP
RazeeDeployment is responsible for creating a RazeeDeploy - Job, and ensuring Razee resources are installed in the Cluster. Currently RazeeDeployment creates:
- ConfigMap -> WatchKeeperNonNamespaced
- ConfigMap -> WatchKeeperLimitPoll
- ConfigMap -> RazeeClusterMetaData
- ConfigMap -> WatchKeeperConfig
- ConfigMap -> WatchKeeperSecret
- Secret -> IbmCosReaderKey
- Job -> RazeeDeployJob
- Job -> RazeeUninstallJob
- RemoteResourceS3
Node controller is responsible for ensuring Kubernetes nodes are patched with the label, razee/watch-resource: lite.
Subscription Controller is responsible for creating operator groups during installation. And deleting CSV's and subscriptions during uninstallation.