Skip to main content
Version: v3.9.0

Microservice templates

A microservice template is a reusable blueprint for one microservice. It stores variables and a microservice skeleton (images, container, config, catalog hooks). It is not a running container.

Use it when many services share the same container shape and differ by Edgelet node, image tag, model name, or environment. For a whole app, use an application template.

Instantiation is spec.template on kind: Microservice. There is no separate instantiate command. Include template in the microservice YAML and run deploy -f.

The primer is Templates. Fields are on Microservice template fields. Console lists them under Catalog and templates.

Variables use the same substitution as application templates. Quote placeholders. Re-deploy the microservice to pick up new values. Existing microservices are not updated when you edit the template.

The stored template omits instance identity (name, uuid, and a nested template). application and agent.name may be templated strings.

What deploy -f does​

potctl deploy -f ms-template.yaml -n my-ecn

deploy parses spec in non-strict mode and creates or updates the template on the Controller. Deploy the same metadata.name again to update.

Examples​

Image and port​

nginx-edge.yaml
apiVersion: datasance.com/v3
kind: MicroserviceTemplate
metadata:
  name: nginx-edge
spec:
  description: nginx with a templated listen port
  variables:
    - key: port
      description: Host port mapped to container port 80
      defaultValue: "80"
    - key: agent-name
      description: Edgelet node name
    - key: app-name
      description: Parent application name
  microservice:
    application: "{{ app-name }}"
    agent:
      name: "{{ agent-name }}"
    images:
      registry: remote
      amd64: nginx:alpine
    container:
      ports:
        - internal: 80
          external: "{{ port }}"
          protocol: tcp

Deploy an instance. Omit variables that match a defaultValue. variables may be a list of {key, value} or a map.

web-zebra-1.yaml
apiVersion: datasance.com/v3
kind: Microservice
metadata:
  name: web-zebra-1
spec:
  application: edge-gateway
  agent:
    name: zebra-1
  template:
    name: nginx-edge
    variables:
      - key: port
        value: "8080"
      - key: agent-name
        value: zebra-1
      - key: app-name
        value: edge-gateway

Fleet model bind​

Attach the model to the Edgelet node before the microservice starts. See AI Model Catalog.

inference-ms.yaml
apiVersion: datasance.com/v3
kind: MicroserviceTemplate
metadata:
  name: inference-ms
  namespace: my-ecn
spec:
  description: Template for inference microservice with fleet model bind
  variables:
    - key: application
      description: Application name
      defaultValue: myapp
    - key: agent-name
      description: Edgelet node
      defaultValue: edge-01
    - key: model1
      description: Fleet model metadata.name
      defaultValue: smollm2-135m
  microservice:
    application: "{{application}}"
    agent:
      name: "{{agent-name}}"
    images:
      registry: remote
      amd64: ghcr.io/datasance/inference:1.0.0
    models:
      bindPath: /models
      permissions: ro
      items:
        - name: "{{model1}}"
    container:
      ports:
        - internal: 8080
          external: 8080
          protocol: tcp

A microservice that points at this template is in Microservices.

Commands​

potctl get microservice-templates -n my-ecn
potctl describe microservice-template inference-ms -n my-ecn
potctl delete microservice-template inference-ms -n my-ecn
Group 3See anything wrong with the document? Help us improve it!

Edit this page on GitHub!