added workaround to handle the faulted node deletion (#27)

* added sleep

Signed-off-by: Marcel Bindseil <[email protected]>


---------

Signed-off-by: Marcel Bindseil <[email protected]>

Author: bindsi

.github/workflows/build-vm.yaml

@@ -205,12 +205,15 @@ jobs:
             --publisher Microsoft.Azure.Extensions \
             --force-update \
             --protected-settings "{\"commandToExecute\": \" \
+                sudo chmod -R 777 /etc/rancher/k3s/k3s.yaml && \
                 export KUBECONFIG=/etc/rancher/k3s/k3s.yaml && \
                 kubectl config use-context default && \
                 kubectl get pods -A -v6 && \
                 az login --service-principal -u ${{ secrets.AZURE_SP_CLIENT_ID }} -p \"\"${{ secrets.AZURE_SP_CLIENT_SECRET }}\"\" --tenant $tenantId && \
                 az extension add --name connectedk8s --yes && \
                 az extension add --name azure-iot-ops --yes && \
+                sleep 1m && \
+                kubectl delete node \$(kubectl get nodes | grep NotReady | awk '{print \$1}') && \
                 az connectedk8s connect --debug -n $clusterName -l $location -g $resourceGroup --subscription $subscriptionId && \
                 az connectedk8s enable-features --debug -n $clusterName -g $resourceGroup --custom-locations-oid \"\"${{ secrets.CUSTOM_LOCATIONS_OBJECT_ID }}\"\" --features cluster-connect custom-locations && \
                 saId=\$(az storage account create -n $saName -g $resourceGroup --enable-hierarchical-namespace -o tsv --query id) && \

vm.sh

@@ -0,0 +1,18 @@
+#!/bin/bash
+
+az vm extension set \
+            --resource-group rg-arc-vmnext \
+            --vm-name arc-vmnext \
+            --name customScript \
+            --publisher Microsoft.Azure.Extensions \
+            --force-update \
+            --protected-settings "{\"commandToExecute\": \" \
+                export KUBECONFIG=/etc/rancher/k3s/k3s.yaml && \
+                notReadyNotes=\$(kubectl get nodes | grep NotReady | awk '{print \$1}' 2>/dev/null) && \
+                len=\${#notReadyNotes} && \
+                echo \"\"not ready\$len\"\" && \
+                if [[ -z \$notReadyNotes ]]; then echo \"\"xxxx\"\" && kubectl delete node \"\"\$notReadyNotes\"\"; fi && \
+                kubectl config use-context default && \
+                kubectl get pods -A -v6 \
+            \"}"
+