Skip to content

Latest commit

 

History

History
121 lines (82 loc) · 2.94 KB

functions.rst

File metadata and controls

121 lines (82 loc) · 2.94 KB

Kanister Functions

Kanister Functions are written in go and are compiled when building the controller. They are referenced by Blueprints phases. A Kanister Function implements the following go interface:

// Func allows custom actions to be executed.
type Func interface {
    Name() string
    Exec(ctx context.Context, args ...string) error
}

Kanister Functions are registered by the return value of Name(), which must be static.

Each phase in a Blueprint executes a Kanister Function. The Func field in a BlueprintPhase is used to lookup a Kanister Function. After BlueprintPhase.Args are rendered, they are passed into the Kanister Function's Exec() method.

Existing Functions

The Kanister controller ships with the following Kanister Functions out-of-the-box that provide integration with Kubernetes:

KubeExec

KubeExec is similar to running

`kubectl exec -it --namespace <NAMESPACE> <POD> -c <CONTAINER> [CMD LIST...]

It requires at least four arguments. The first three arguments are used to determine the container to exec into. The remaining arguments are grouped and executed as a command.

The arguments are:

  1. namespace
  2. pod
  3. container
  4. [command]

KubeExecAll

KubeExecAll is similar to running KubeExec on multiple containers on multiple pods in parallel.

It requires at least four arguments. The first three arguments are used to determine one or more containers to exec into. The remaining arguments are grouped and executed as a command.

The arguments are:

  1. namespace
  2. pod(s)
  3. container(s)
  4. [command]

KubeTask

KubeTask spins up a new container and executes a command via a Kubernetes job. This allows you to run a new Pod from a Blueprint.

KubeTask takes the following three arguments:

  1. namespace
  2. image
  3. [command]

ScaleDeployment

ScaleDeployment is used to scale up or scale down a deployment. It is similar to running

`kubectl scale deployment <DEPLOYMENT-NAME> --replicas=<NUMBER OF REPLICAS> --namespace <NAMESPACE>

It requires the following three arguments:

  1. namespace
  2. deployment name
  3. number of replicas

ScaleStatefulSet

ScaleStatefulSet is used to scale up or scale down a stateful-set. It is similar to running

`kubectl scale statefulsets <STATEFUL-SET-NAME> --replicas=<NUMBER OF REPLICAS> --namespace <NAMESPACE>

It requires the following three arguments:

  1. namespace
  2. statefulset name
  3. number of replicas

Registering Functions

Kanister can be extended by registering new Kanister Functions.

Kanister Functions are registered using a similar mechanism to database/sql drivers. To register new Kanister Functions, import a package with those new functions into the controller and recompile it.