Merge pull request #18 from ustuzhanin/helm

Helm update

README.md

@@ -61,6 +61,19 @@ helm install --set seaweedfsFiler=<filerHost:port> seaweedfs-csi-driver ./seawee
 ```bash
 helm uninstall seaweedfs-csi-driver
 ```
+
+# Safe rollout update
+When update DaemonSet ( DS ) break processes who implements fuse mount. 
+And now new pod not remount net device
+
+For better safe update use ``node.updateStrategy.type: OnDelete`` in this need manual update. Steps:
+ - delete DS pods on node where no exists seaweefs PV
+ - cordon or taint node 
+ - evict or delete pods with seaweedfs PV
+ - delete DS pod on node
+ - uncordon or remove taint on node
+ - repeat all steps on all nodes 
+  
 # License
 [Apache v2 license](https://www.apache.org/licenses/LICENSE-2.0)
 

deploy/helm/seaweedfs-csi-driver/.helmignore

@@ -21,3 +21,4 @@
 .idea/
 *.tmproj
 .vscode/
+.tgz

deploy/helm/seaweedfs-csi-driver/Chart.yaml

@@ -2,5 +2,5 @@ apiVersion: v2
 name: seaweedfs-csi-driver
 description: A Helm chart for Kubernetes
 type: application
-version: 0.1.0
+version: 0.1.1
 appVersion: latest

deploy/helm/seaweedfs-csi-driver/templates/daemonset.yml

@@ -9,8 +9,7 @@ spec:
     matchLabels:
       app: {{ template "seaweedfs-csi-driver.name" . }}-node
   updateStrategy:
-    rollingUpdate:
+    {{ toYaml .Values.node.updateStrategy | nindent 4 }}
-      maxUnavailable: 25%
   template:
     metadata:
       labels:

deploy/helm/seaweedfs-csi-driver/values.yaml

@@ -28,5 +28,12 @@ driverName: seaweedfs-csi-driver
 
 node:
   # Deploy node daemonset
-  # for a smoother update it is better to use static pods on nodes
   enabled: true
+  # When pod on node be recreated all pod on same node lost PV.
+  # For safe update use updateStrategy.type: OnDelete and manual move pods who use PV and delete damonset pod
+  updateStrategy:
+    type: RollingUpdate
+    rollingUpdate:
+      maxUnavailable: 25%
+
+