Manage Definition

Manage Definition - 图1tip

Before reading this part, please make sure you’ve learned the Definition Concept of KubeVela.

In KubeVela CLI, vela def command group provides a series of convenient definition writing tools. With these commands, users only need to write CUE files to generate and edit definitions, instead of composing Kubernetes YAML object with mixed CUE string.

init

vela def init is a command that helps users bootstrap new definitions. To create an empty trait definition, run:

  1. vela def init my-trait -t trait --desc "My trait description."

It will generate the following scaffold:

  1. "my-trait": {
  2.         annotations: {}
  3.         attributes: {
  4.                 appliesToWorkloads: []
  5.                 conflictsWith: []
  6.                 definitionRef:   ""
  7.                 podDisruptive:   false
  8.                 workloadRefPath: ""
  9.         }
  10.         description: "My trait description."
  11.         labels: {}
  12.         type: "trait"
  13. }
  14. template: patch: {}

You can also initiate definitions interactively like below:

  1. vela def init my-comp --interactive

expected output

  1. Please choose one definition type from the following values: component, trait, policy, workload, scope, workflow-step
  2. > Definition type: component
  3. > Definition description: My component definition.
  4. Please enter the location the template YAML file to build definition. Leave it empty to generate default template.
  5. > Definition template filename: 
  6. Please enter the output location of the generated definition. Leave it empty to print definition to stdout.
  7. > Definition output filename: my-component.cue
  8. Definition written to my-component.cue

In addition, users can create definitions from existing YAML files. For example, if a user want to create a ComponentDefinition which is designed to generate a deployment, and this deployment has already been created elsewhere, he/she can use the --template-yaml flag to complete the transformation. The YAML file is as below

  1. # my-deployment.yaml
  2. apiVersion: apps/v1
  3. kind: Deployment
  4. metadata:
  5.   name: hello-world
  6. spec:
  7.   replicas: 1
  8.   selector:
  9.     matchLabels:
  10.       app.kubernetes.io/name: hello-world
  11.   template:
  12.     metadata:
  13.       labels:
  14.         app.kubernetes.io/name: hello-world
  15.     spec:
  16.       containers:
  17.       - name: hello-world
  18.         image: somefive/hello-world
  19.         ports: 
  20.         - name: http
  21.           containerPort: 80
  22.           protocol: TCP
  23. ---
  24. apiVersion: v1
  25. kind: Service
  26. metadata:
  27.   name: hello-world-service
  28. spec:
  29.   selector:
  30.     app: hello-world
  31.   ports:
  32.   - name: http
  33.     protocol: TCP
  34.     port: 80
  35.     targetPort: 8080
  36.   type: LoadBalancer

Running the following command to get the CUE-format ComponentDefinition:

  1. vela def init my-comp -t component --desc "My component." --template-yaml ./my-deployment.yaml

The expect output will be:

  1. "my-comp": {
  2.         annotations: {}
  3.         attributes: workload: definition: {
  4.                 apiVersion: "<change me> apps/v1"
  5.                 kind:       "<change me> Deployment"
  6.         }
  7.         description: "My component."
  8.         labels: {}
  9.         type: "component"
  10. }
  11. template: {
  12.         output: {
  13.                 metadata: name: "hello-world"
  14.                 spec: {
  15.                         replicas: 1
  16.                         selector: matchLabels: "app.kubernetes.io/name": "hello-world"
  17.                         template: {
  18.                                 metadata: labels: "app.kubernetes.io/name": "hello-world"
  19.                                 spec: containers: [{
  20.                                         name:  "hello-world"
  21.                                         image: "somefive/hello-world"
  22.                                         ports: [{
  23.                                                 name:          "http"
  24.                                                 containerPort: 80
  25.                                                 protocol:      "TCP"
  26.                                         }]
  27.                                 }]
  28.                         }
  29.                 }
  30.                 apiVersion: "apps/v1"
  31.                 kind:       "Deployment"
  32.         }
  33.         outputs: "hello-world-service": {
  34.                 metadata: name: "hello-world-service"
  35.                 spec: {
  36.                         ports: [{
  37.                                 name:       "http"
  38.                                 protocol:   "TCP"
  39.                                 port:       80
  40.                                 targetPort: 8080
  41.                         }]
  42.                         selector: app: "hello-world"
  43.                         type: "LoadBalancer"
  44.                 }
  45.                 apiVersion: "v1"
  46.                 kind:       "Service"
  47.         }
  48.         parameter: {}
  50. }

Then the user can make further modifications based on the definition file above, like removing <change me> in workload.definition。

vet

After initializing definition files, run vela def vet <my-def.cue> to validate if there are any syntax error in the definition file.

  1. vela def vet my-comp.cue

It can be used to detect some simple errors such as missing brackets.

apply

After confirming the definition file has correct syntax, users can apply this definition into control plane by:

  1. vela def apply my-comp.cue --namespace my-namespace

The definition will be applied to the namespace named my-namespace, by default the namespace is vela-system。

Manage Definition - 图2tip

When definition applied to vela-system, the definition can be used for all applications in this cluster. If definitions are applied in some other namespaces, only the application in the same namespace can use that. It’s useful when in multi-tenant cases.

dry-run

If you want to check the transformed Kubernetes API format, you can use --dry-run can achieve that.

  1. vela def apply my-comp.cue --dry-run

expected output

  1. apiVersion: core.oam.dev/v1beta1
  2. kind: ComponentDefinition
  3. metadata:
  4.   annotations:
  5.     definition.oam.dev/description: My component.
  6.   labels: {}
  7.   name: my-comp
  8.   namespace: vela-system
  9. spec:
  10.   schematic:
  11.     cue:
  12.       template: |
  13.         output: {
  14.                 metadata: name: "hello-world"
  15.                 spec: {
  16.                         replicas: 1
  17.                         selector: matchLabels: "app.kubernetes.io/name": "hello-world"
  18.                         template: {
  19.                                 metadata: labels: "app.kubernetes.io/name": "hello-world"
  20.                                 spec: containers: [{
  21.                                         name:  "hello-world"
  22.                                         image: "somefive/hello-world"
  23.                                         ports: [{
  24.                                                 name:          "http"
  25.                                                 containerPort: 80
  26.                                                 protocol:      "TCP"
  27.                                         }]
  28.                                 }]
  29.                         }
  30.                 }
  31.                 apiVersion: "apps/v11"
  32.                 kind:       "Deployment"
  33.         }
  34.         outputs: "hello-world-service": {
  35.                 metadata: name: "hello-world-service"
  36.                 spec: {
  37.                         ports: [{
  38.                                 name:       "http"
  39.                                 protocol:   "TCP"
  40.                                 port:       80
  41.                                 targetPort: 8080
  42.                         }]
  43.                         selector: app: "hello-world"
  44.                         type: "LoadBalancer"
  45.                 }
  46.                 apiVersion: "v1"
  47.                 kind:       "Service"
  48.         }
  49.         parameter: {}
  50.   workload:
  51.     definition:
  52.       apiVersion: apps/v1
  53.       kind: Deployment

get

While you can use native kubectl tools to confirm the results of the apply command, as mentioned above, the YAML object mixed with raw CUE template string is complex. Using vela def get will automatically convert the YAML object into the CUE-format definition.

  1. vela def get my-comp

list

You can list all definitions installed through:

  1. vela def list

You can filter type by:

  • List component only:

    1. vela def list -t component

  • List trait only:

    1. vela def list -t trait

  • List workflow step only:

    1. vela def list -t workflow-step

  • List policy only:

    1. vela def list -t policy

edit

Using vela def edit to edit definitions in pure CUE-format. The transformation between CUE-format definition and YAML object is done by the command. Besides, you can specify the EDITOR environment variable to use your favorite editor.

  1. EDITOR=vim vela def edit my-comp

delete

vela def del can be utilized to delete existing definitions.

  1. $ vela def del my-comp -n my-namespace  
  2. Are you sure to delete the following definition in namespace my-namespace?
  3. ComponentDefinition my-comp: My component.
  4. [yes|no] > yes
  5. ComponentDefinition my-comp in namespace my-namespace deleted.

A brief quick view for these commands are shown below.

def-demo

debug with applications

When we want to debug definition with application, we can use vela dry-run --definitions or -d for short, to specify local definitions.

Dry run will help you to understand what are the real resources which will to be expanded and deployed to the Kubernetes cluster. In other words, it will mock to run the same logic as KubeVela’s controller and output the results locally.

For example, let’s dry-run the following application:

  1. # app.yaml
  2. apiVersion: core.oam.dev/v1beta1
  3. kind: Application
  4. metadata:
  5.   name: vela-app
  6. spec:
  7.   components:
  8.     - name: express-server
  9.       type: my-comp

Use dry run for test:

  1. vela dry-run -f app.yaml -d my-comp.cue

Manage Definition - 图4caution

Before KubeVela CLI 1.6.2, you can only use definition in Kubernetes API format.

  1. vela dry-run -f app.yaml -d my-comp.yaml

expected output

  1. ---
  2. # Application(vela-app) -- Component(express-server)
  3. ---
  5. apiVersion: apps/v1
  6. kind: Deployment
  7. metadata:
  8.   annotations: {}
  9.   labels:
  10.     app.oam.dev/appRevision: ""
  11.     app.oam.dev/component: express-server
  12.     app.oam.dev/name: vela-app
  13.     app.oam.dev/namespace: default
  14.     app.oam.dev/resourceType: WORKLOAD
  15.     workload.oam.dev/type: my-comp
  16.   name: hello-world
  17.   namespace: default
  18. spec:
  19.   replicas: 1
  20.   selector:
  21.     matchLabels:
  22.       app.kubernetes.io/name: hello-world
  23.   template:
  24.     metadata:
  25.       labels:
  26.         app.kubernetes.io/name: hello-world
  27.     spec:
  28.       containers:
  29.       - image: somefive/hello-world
  30.         name: hello-world
  31.         ports:
  32.         - containerPort: 80
  33.           name: http
  34.           protocol: TCP
  36. ---
  37. ## From the auxiliary workload
  38. apiVersion: v1
  39. kind: Service
  40. metadata:
  41.   annotations: {}
  42.   labels:
  43.     app.oam.dev/appRevision: ""
  44.     app.oam.dev/component: express-server
  45.     app.oam.dev/name: vela-app
  46.     app.oam.dev/namespace: default
  47.     app.oam.dev/resourceType: TRAIT
  48.     trait.oam.dev/resource: hello-world-service
  49.     trait.oam.dev/type: AuxiliaryWorkload
  50.   name: hello-world-service
  51.   namespace: default
  52. spec:
  53.   ports:
  54.   - name: http
  55.     port: 80
  56.     protocol: TCP
  57.     targetPort: 8080
  58.   selector:
  59.     app: hello-world
  60.   type: LoadBalancer
  62. ---

Next Step