StorageOS plugin for managed Kubernetes service

When StorageOS is used on a managed platform such as OVH Managed Kubernetes Service, the end user doesn't have access the cluster node.

The cluster rolling upgrade is done by the service provider. Once the node N is Ready, it'll update the node N+1 if in Ready state.

By default, the update process is unaware of higher level of service deployed by the user.

To keep data consistency, the end-user must signal to the update process whether or not the node can be updated.

This is the goal of this plugin.

Prepare your environment

First thing is to setup your environment. I usually put all that part in a activate file. Edit the file according to our setting.

. ./activate

Install StorageOS

Follow the instruction here https://docs.storageos.com/docs/platforms/kubernetes/install/1.15

Note: you must use an external etcd

or with helm

helm repo add storageos https://charts.storageos.com
helm install \
  --set cluster.namespace=storageos \
  --set cluster.secretRefName=storageos-api \
  --set cluster.kvBackend.embedded=false \
  --set cluster.kvBackend.address=http://<ETCD_SERVER>:2379 \
  --set cluster.kvBackend.backend=etcd \
  --set cluster.csi.enable=true \
  --namespace storageos-operator \
  --name storageos \
  storageos/storageos-operator

(How to install helm: https://gist.github.com/ticapix/762878ad070fcc1d164ff35fbc25b5ca)

Deploy the service

The DaemonSet can be deployed and undeployed with make.

make deploy
# or
make undeploy

TODO: use helm instead