Deploying registries in Kubernetes | Eclipse Che Documentation

Procedure

A Helm chart for the plug-in registry is available in the /kubernetes/che-plugin-registry/ directory of the GitHub repository.

To deploy the plug-in registry using the Helm chart, run the following command:

NAMESPACE=<namespace-name>     (1)
DOMAIN=<kubernetes-cluster-domain> (2)
IMAGE="my-plugin-registry"
helm upgrade --install che-plugin-registry \
 --debug \
 --namespace $\{NAMESPACE} \
 --set global.ingressDomain=$\{DOMAIN} \
 --set chePluginRegistryImage=$\{IMAGE} \
 --set chePluginRegistryImagePullPolicy="IfNotPresent" \
 ./kubernetes/che-plugin-registry/

1	The default Che namespace is `che`.
2	On Minikube, use `$(minikube ip).nip.io`

The devfile registry also has a Helm chart in the deploy/kubernetes/che-devfile-registry/ directory of the GitHub repository. To deploy it, run the command:

NAMESPACE=<namespace-name>     (1)
DOMAIN=<kubernetes-cluster-domain> (2)
IMAGE="my-devfile-registry"
helm upgrade --install che-devfile-registry \
 --debug \
 --namespace $\{NAMESPACE} \
 --set global.ingressDomain=$\{DOMAIN} \
 --set cheDevfileRegistryImage=$\{IMAGE} \
 --set cheDevfileRegistryImagePullPolicy="IfNotPresent" \
 ./deploy/kubernetes/che-devfile-registry/

1	The default Che namespace is `che`.
2	On Minikube, use `$(minikube ip).nip.io`

The Helm chart creates a pod, a service, and an Ingress. To get them, use app=che-plugin-registry (or app=che-plugin-registry for the devfile registry).

$ kubectl get -o custom-columns=TYPE:.kind,NAME:.metadata.name \
  -l app=che-plugin-registry pod,svc,ingress
TYPE      NAME
Pod       che-plugin-registry-5c7cd8d5c9-zlqlz
Service   che-plugin-registry
Ingress   che-plugin-registry

To verify that the new plug-in is correctly published to the plug-in registry, make a request to the registry path /v3/plugins/index.json (or /devfiles/index.json for the devfile registry).

$ URL=$(kubectl get -o 'custom-columns=URL:.spec.rules[0].host' \
  -l app=che-plugin-registry ingress --no-headers)
$ INDEX_JSON=$(curl -sSL http://$\{URL}/v3/plugins/index.json)
$ echo $\{INDEX_JSON} | grep -A 4 -B 5 "\"name\":\"my-plugin\""
,\{
 "id": "my-org/my-plugin/1.0.0",
 "displayName":"This is my first Che plugin",
 "version":"1.0.0",
 "type":"VS Code extension",
 "name":"my-plugin",
 "description":"This plugins shows that we are able to add plugins to the registry...",
 "publisher":"my-org",
 "links": \{"self":"/v3/plugins/my-org/my-plugin/1.0.0" }
}
--
--
,\{
 "id": "my-org/my-plugin/latest",
 "displayName":"This is my first Che plugin",
 "version":"latest",
 "type":"VS Code extension",
 "name":"my-plugin",
 "description":"This plugins shows that we are able to add plugins to the registry...",
 "publisher":"my-org",
 "links": \{"self":"/v3/plugins/my-org/my-plugin/latest" }
}

Verify that the Che server points to the URL of the registry. To do this, compare the value of the CHE_WORKSPACE_PLUGIN__REGISTRY__URL parameter in the che ConfigMap (or CHE_WORKSPACE_DEVFILE__REGISTRY__URL for the devfile registry):

$ kubectl get \
  -o "custom-columns=URL:.data['CHE_WORKSPACE_PLUGIN__REGISTRY__URL']" \
  --no-headers cm/che
URL
http://che-plugin-registry-che.192.168.99.100.nip.io/v3

with the URL of the Ingress:

$ kubectl get -o 'custom-columns=URL:.spec.rules[0].host' \
  -l app=che-plugin-registry ingress --no-headers
che-plugin-registry-che.192.168.99.100.nip.io

If they do not match, update the ConfigMap and restart the Che server.
```
$ kubectl edit cm/che
(...)
$ kubectl scale --replicas=0 deployment/che
$ kubectl scale --replicas=1 deployment/che
```
When the new registries are deployed and the Che server is configured to use them, the new plug-ins are available in the Plugin view of a workspace.

The new stacks are displayed in the New Workspace tab of the user dashboard.