Change container/k8s resources to a consistent name schema #53
Labels
enhancement
New feature or request
Comments
fatih
added a commit
that referenced
this issue
Aug 31, 2018
This PR does the following: * combine the `csi-provisioner` and `csi-attacher` into a single `csi-do-controller` statefulset * rename the container name inside the `attacher` and `provisioner` statefulsets from `digitalocean-csi-plugin` to `csi-do-plugin` * rename the container name inside the `node` daemonset from `csi-doplugin` to `csi-do-plugin` * remove the RBAC's for the external attacher and provisioner as they are now included starting with K8S v1.10.5 * renamed the service account names to be consistent with the overall naming * removed the `Service` resources as those are not needed * changed the RBAC permission for the `node` plugin. It was too permissive (initially intended for the controller sidecars). * added a common label called `role=csi-do` to all pods, and `app=csi-do-controller` and `app=csi-do-node` to the controller and node plugin This allows you to get the pods or logs easily. Some example commands: ``` $ kubectl get pods -n kube-system NAME READY STATUS RESTARTS AGE csi-do-controller-0 3/3 Running 0 15s csi-do-node-4tkvh 2/2 Running 0 14s csi-do-node-4vjpr 2/2 Running 0 14s csi-do-node-thvjc 2/2 Running 0 14s kube-dns-64f766c69c-qsdnz 3/3 Running 0 49m kube-proxy-worker-7871 1/1 Running 1 48m kube-proxy-worker-7872 1/1 Running 1 49m kube-proxy-worker-7873 1/1 Running 1 48m kubernetes-dashboard-7dd4fc69c8-fs8s5 1/1 Running 0 49m $ kubectl get pods -l app=csi-do-controller -n kube-system NAME READY STATUS RESTARTS AGE csi-do-controller-0 3/3 Running 0 34s $ kubectl get pods -l app=csi-do-node -n kube-system NAME READY STATUS RESTARTS AGE csi-do-node-4tkvh 2/2 Running 0 39s csi-do-node-4vjpr 2/2 Running 0 39s csi-do-node-thvjc 2/2 Running 0 39s $ kubectl get pods -l role=csi-do -n kube-system NAME READY STATUS RESTARTS AGE csi-do-controller-0 3/3 Running 0 49s csi-do-node-4tkvh 2/2 Running 0 48s csi-do-node-4vjpr 2/2 Running 0 48s csi-do-node-thvjc 2/2 Running 0 48s ``` closes #53
fatih
added a commit
that referenced
this issue
Aug 31, 2018
This PR does the following: * combine the `csi-provisioner` and `csi-attacher` into a single `csi-do-controller` statefulset * rename the container name inside the `attacher` and `provisioner` statefulsets from `digitalocean-csi-plugin` to `csi-do-plugin` * rename the container name inside the `node` daemonset from `csi-doplugin` to `csi-do-plugin` * remove the RBAC's for the external attacher and provisioner as they are now included starting with K8S v1.10.5 * renamed the service account names to be consistent with the overall naming * removed the `Service` resources as those are not needed * changed the RBAC permission for the `node` plugin. It was too permissive (initially intended for the controller sidecars). * added a common label called `role=csi-do` to all pods, and `app=csi-do-controller` and `app=csi-do-node` to the controller and node plugin This allows you to get the pods or logs easily. Some example commands: ``` $ kubectl get pods -n kube-system NAME READY STATUS RESTARTS AGE csi-do-controller-0 3/3 Running 0 15s csi-do-node-4tkvh 2/2 Running 0 14s csi-do-node-4vjpr 2/2 Running 0 14s csi-do-node-thvjc 2/2 Running 0 14s kube-dns-64f766c69c-qsdnz 3/3 Running 0 49m kube-proxy-worker-7871 1/1 Running 1 48m kube-proxy-worker-7872 1/1 Running 1 49m kube-proxy-worker-7873 1/1 Running 1 48m kubernetes-dashboard-7dd4fc69c8-fs8s5 1/1 Running 0 49m $ kubectl get pods -l app=csi-do-controller -n kube-system NAME READY STATUS RESTARTS AGE csi-do-controller-0 3/3 Running 0 34s $ kubectl get pods -l app=csi-do-node -n kube-system NAME READY STATUS RESTARTS AGE csi-do-node-4tkvh 2/2 Running 0 39s csi-do-node-4vjpr 2/2 Running 0 39s csi-do-node-thvjc 2/2 Running 0 39s $ kubectl get pods -l role=csi-do -n kube-system NAME READY STATUS RESTARTS AGE csi-do-controller-0 3/3 Running 0 49s csi-do-node-4tkvh 2/2 Running 0 48s csi-do-node-4vjpr 2/2 Running 0 48s csi-do-node-thvjc 2/2 Running 0 48s ``` closes #53
fatih
added a commit
that referenced
this issue
Sep 5, 2018
This PR does the following: * combine the `csi-provisioner` and `csi-attacher` into a single `csi-do-controller` statefulset * rename the container name inside the `attacher` and `provisioner` statefulsets from `digitalocean-csi-plugin` to `csi-do-plugin` * rename the container name inside the `node` daemonset from `csi-doplugin` to `csi-do-plugin` * remove the RBAC's for the external attacher and provisioner as they are now included starting with K8S v1.10.5 * renamed the service account names to be consistent with the overall naming * removed the `Service` resources as those are not needed * changed the RBAC permission for the `node` plugin. It was too permissive (initially intended for the controller sidecars). * added a common label called `role=csi-do` to all pods, and `app=csi-do-controller` and `app=csi-do-node` to the controller and node plugin This allows you to get the pods or logs easily. Some example commands: ``` $ kubectl get pods -n kube-system NAME READY STATUS RESTARTS AGE csi-do-controller-0 3/3 Running 0 15s csi-do-node-4tkvh 2/2 Running 0 14s csi-do-node-4vjpr 2/2 Running 0 14s csi-do-node-thvjc 2/2 Running 0 14s kube-dns-64f766c69c-qsdnz 3/3 Running 0 49m kube-proxy-worker-7871 1/1 Running 1 48m kube-proxy-worker-7872 1/1 Running 1 49m kube-proxy-worker-7873 1/1 Running 1 48m kubernetes-dashboard-7dd4fc69c8-fs8s5 1/1 Running 0 49m $ kubectl get pods -l app=csi-do-controller -n kube-system NAME READY STATUS RESTARTS AGE csi-do-controller-0 3/3 Running 0 34s $ kubectl get pods -l app=csi-do-node -n kube-system NAME READY STATUS RESTARTS AGE csi-do-node-4tkvh 2/2 Running 0 39s csi-do-node-4vjpr 2/2 Running 0 39s csi-do-node-thvjc 2/2 Running 0 39s $ kubectl get pods -l role=csi-do -n kube-system NAME READY STATUS RESTARTS AGE csi-do-controller-0 3/3 Running 0 49s csi-do-node-4tkvh 2/2 Running 0 48s csi-do-node-4vjpr 2/2 Running 0 48s csi-do-node-thvjc 2/2 Running 0 48s ``` closes #53
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Daemonset name:
csi-dopluginContainer name:
csi-dopluginStatefulset name:
csi-attacher-dopluginContainer name:
digitalocean-csi-pluginStatefulset name:
csi-provisioner-dopluginContainer name:
digitalocean-csi-pluginOur image name:
digitalocean/do-csi-pluginCurrently these are the names we're using and it's confusing. We should name them consistently. Suggestions:
csi-node-plugin(daemonset), the container inside the pod will be called:csi-plugincsi-controller-plugin(statefulset)csi-attacher-dopluginandcsi-provisionerplugin into a single statefulsetcsi-plugindigitalocean/csi-pluginThis is not the final decision, please comment if you have concerns or any suggestions.
The text was updated successfully, but these errors were encountered: