Data sharing controllers on Kubernetes

Overview

Data sharing controllers manage shared data, i.e. provided data, according to sharing policies, on single and multiple Kubernetes clusters. This enables data providers and consumers to release from the cumbersome and error-prone management tasks. The controllers present the following three functionalities.

1. Updating continuously copied provided data and deleting them at the expiry date.

2. Deleting derived data as well as origins of provided data

3. Controlling accesses to provided data at the granularity of application programs

Requirements

• Kubernetes
  • Custom resource definitions for the data lifetime controller are registered in a Kubernetes cluster.
• Docker engine
  • Client code for Kubernetes custom resource APIs is generated by the code generator in a docker container.
  • Client code for gRPC APIs is generated in a docker container.
  • Test code is run in a docker container.
• Docker registry
  • The docker images including generated executables, the data lifetime controller, the volume controller and its client, and the lifetime validator are stored in this registry.
    Set the environment variable ${DOCKER_REGISTRY} if you use your private docker registry.
• Apache Kafka
  • Shared data are transferred from providers to consumers through message queues.

Creating quickly the docker images including data sharing controllers

$ make

Creating the docker images including the controllers using customized Kubernetes custom resource code

$ make cr-codegen
$ make

Give the ${CODEGEN_BUILD_CACHE} environment variable if you would like to generate the docker image including the client code generator only the first time this command is issued.

$ CODEGEN_BUILD_CACHE=true make cr-codegen
$ make

Creating the docker images including the controllers using customized gRPC code

$ make grpc-codegen
$ make

Give the ${CODEGEN_BUILD_CACHE} environment variable if you would like to generate the docker image including the client code generator only the first time this command is issued.

$ CODEGEN_BUILD_CACHE=true make grpc-codegen
$ make

Creating CSI driver images for the volume controller

Refer to the artifacts/csi_drivers directory.

Running the controllers on Kubernetes

Refer to the steps until the data sharing controllers start managing a data lifetime using example data under the artifacts/examples directory.

Carrying out test cases for the source code

The Kafka instance is required to run test cases. You need to launch the containerized instance provided by the kafka-go library in advance as described here.

docker-compose up -d

Use the variable ${TFLAGS} if you would like to give go test arguments, such as -v and -short, as follows.

$ TFLAGS="[test flags]" make test

Use the variable ${GO_TEST} if you would like to run each test case as follows.

$ GO_TEST="go test ([build/test flags]) [package name] -run [test name]" make test

Cleaning up the source code tree

Run the following command to delete the executables.

$ make clean

Run the following command to delete all the generated files including the executables, the client codes generated by the code generators, and docker images.

$ make distclean

Contact

contact-dscok@cs.jp.fujitsu.com