Added Helm charts for deployment. #35
Conversation
| registrar: quay.io/k8scsi/csi-node-driver-registrar:v2.0.1 | ||
| huaweiCsiControllerService: huawei-csi:test | ||
| huaweiCsiNodeService: huawei-csi:test | ||
| imagePullPolicy: "IfNotPresent" |
There was a problem hiding this comment.
Please add comments on each attribute for easy of use
| name: huawei-csi-configmap | ||
| - name: secret | ||
| secret: | ||
| secretName: huawei-csi-secret |
| - name: registration-dir | ||
| mountPath: /registration | ||
| - name: huawei-csi-driver | ||
| image: {{ required "Must provide the CSI controller service node image." .Values.images.huaweiCsiNodeService }} |
| name: huawei-csi-configmap | ||
| - name: secret | ||
| secret: | ||
| secretName: huawei-csi-secret |
| capabilities: | ||
| add: ["SYS_ADMIN"] | ||
| allowPrivilegeEscalation: true | ||
| imagePullPolicy: "IfNotPresent" |
There was a problem hiding this comment.
Please add image pull policy template here
| valueFrom: | ||
| fieldRef: | ||
| apiVersion: v1 | ||
| fieldPath: spec.nodeName |
There was a problem hiding this comment.
IMO, Currently the "CSI_ESDK_NODENAME" is not used. Please remove it
There was a problem hiding this comment.
Should I leave the env: section empty or remove the section altogether?
There was a problem hiding this comment.
I have removed the env variable CSI_ESDK_NODENAME for now and retained the env: section.
…d new line at eof
| namespace: kube-system | ||
|
|
||
| --- | ||
| kind: ClusterRole |
There was a problem hiding this comment.
According to Helm best practices each resource definition should be defined in a separate template.
| metadata: | ||
| name: huawei-csi-controller | ||
| namespace: kube-system | ||
| --- |
There was a problem hiding this comment.
According to Helm best practices each resource definition should be defined in a separate template.
| apiVersion: apps/v1 | ||
| metadata: | ||
| name: huawei-csi-controller | ||
| namespace: kube-system |
There was a problem hiding this comment.
Hardcoding a namespace is not recommended in Helm charts. We should discuss how to best approach the issue with the kube-system namespace being hardcoded in secretGenerate and secretUpdate.
| template: | ||
| metadata: | ||
| labels: | ||
| app: huawei-csi-controller |
There was a problem hiding this comment.
Using a the default template function for labels created by helm create ... would allow you to be more consistent regarding your labels and simplify the configuration of labels. See ... for how we're using the .labels and .selectorLabels from _helpers.tpl in our Helm chart: https://github.com/adfinis-sygroup/helm-charts/blob/c23bb7d8380df256e454cc143510de60511ae2e2/charts/huawei-csi-plugin/templates/daemonset-node.yaml#L7
| data: | ||
| csi.json: | | ||
| {{ $length := len .Values.backends }} {{ if gt $length 0 }} { {{ end }} | ||
| "backends": {{ .Values.backends | toPrettyJson | nindent 8 }} |
There was a problem hiding this comment.
In our chart we decided to just dump the YAML structure from values.yaml directly into csi.json: https://github.com/adfinis-sygroup/helm-charts/blob/c23bb7d8380df256e454cc143510de60511ae2e2/charts/huawei-csi-plugin/templates/configmap.yaml#L9
Might be easier then constructing a JSON and more flexible if you would decide to add more then just the backends section in the future.
Added Helm3 support for deploying Huawei CSI driver for Kubernetes.
values.yaml