Discovery service

Discovery service is used by StorageOS to help forming clusters.

API reference

Create new cluster

Creates new cluster:

curl --request POST \
  --url https://discovery.storageos.cloud/clusters

Response:

{
	"id": "8976384d-08c3-4c3a-b3a9-5e3a6def7062", 
	"size": 3, 
	"createdAt": "2017-07-19T14:08:59.724988221Z",
	"updatedAt": "2017-07-19T14:08:59.724988559Z"
}

Here:

• id - cluster ID, should be supplied to StorageOS through env variable CLUSTER_ID
• size - expected cluster size, StorageOS will wait for 3 members to register before starting

Get cluster status

Get cluster status (expected size, creation date and registered member info):

curl --request GET \
  --url https://discovery.storageos.cloud/clusters/8976384d-08c3-4c3a-b3a9-5e3a6def7062

Response:

{
	"id": "8976384d-08c3-4c3a-b3a9-5e3a6def7062",
	"size": 3,
	"nodes": [
		{
			"id": "node-id",
			"name": "storageos-1",
			"advertiseAddress": "http://1.1.1.1:2380",
			"createdAt": "2017-07-19T14:13:29.182503707Z",
			"updatedAt": "2017-07-19T14:13:29.182503807Z"
		}
	],
	"createdAt": "2017-07-19T14:08:59.724988221Z",
	"updatedAt": "2017-07-19T14:08:59.724988559Z"
}

Register node (internal, used by StorageOS)

StorageOS is using this API for node registration but in some cases it can be useful for debugging:

curl --request PUT \
  --url https://discovery.storageos.cloud/clusters/8976384d-08c3-4c3a-b3a9-5e3a6def7062 \
  --header 'content-type: application/json' \
  --data '{\n	"id": "node-id",\n	"name":"storageos-1",\n	"advertiseAddress": "http://1.1.1.1:2380"\n}'

Response is same as status API call.

Building an image

There is a cloudbuild.yaml for Google Cloud Container Builder that can work for you with little modification (project ID). But recommended solution is to use multi-stage Dockerfile:

docker build -t <your org name>/discovery:latest -f Dockerfile.multi .

Deployment on Kubernetes

Discovery wants persistent storage for storing cluster related data, create gcloud disk:

gcloud compute disks create --size 10GB cluster-db

Then, use supplied hack/deployment.yml file to create a deployment:

kubectl create -f hack/deployment.yml

Create a service to expose it internally on a NodePort:

kubectl create -f hack/svc.yml    

Expose through an ingress (the ingress resource provided is a partial example, you'll have to configure it how you require):

kubectl create -f hack/ingress.yml