TopoLVM needs startup buffer to wait on dns pod startup

src/topolvm/assets/03-topolvm.yaml

@@ -865,6 +865,13 @@ spec:
               mountPropagation: "Bidirectional"
             - name: devices-dir
               mountPath: /dev
+          startupProbe:
+            failureThreshold: 60
+            periodSeconds: 2
+            timeoutSeconds: 3
+            httpGet:
+              port: healthz
+              path: /healthz
         - name: csi-registrar
           image: "ghcr.io/topolvm/topolvm-with-sidecar:0.36.2"
           command:
@@ -984,18 +991,29 @@ spec:
               path: /readyz
               port: readyz
               scheme: HTTP
+            timeoutSeconds: 3
+            failureThreshold: 3
+            periodSeconds: 60
           livenessProbe:
             httpGet:
               path: /healthz
               port: healthz
             initialDelaySeconds: 10
             timeoutSeconds: 3
             periodSeconds: 60
+            failureThreshold: 3
           volumeMounts:
             - name: socket-dir
               mountPath: /run/topolvm
             - name: certs
               mountPath: /certs
+          startupProbe:
+            failureThreshold: 3
+            periodSeconds: 60
+            timeoutSeconds: 3
+            httpGet:
+              port: healthz
+              path: /healthz
         - name: csi-provisioner
           image: "ghcr.io/topolvm/topolvm-with-sidecar:0.36.2"
           command:

src/topolvm/generate_manifests.sh

@@ -37,20 +37,40 @@ EOF
   # Patch replicas to 1
   # shellcheck disable=SC2016
   yq 'select(.kind == "Deployment").spec.replicas = 1' -i "${ASSETS_DIR}/03-topolvm.yaml"
+
+  # Patch topolvm-controller manifest with longer startup delay to allow dns to start
+  yq 'with(select(.kind == "Deployment" and .metadata.name == "topolvm-controller").spec.template.spec.containers[] | select(.name == "topolvm-controller");
+  .livenessProbe.failureThreshold = 3 |
+  .readinessProbe.timeoutSeconds = 3 |
+  .readinessProbe.failureThreshold = 3 |
+  .readinessProbe.periodSeconds = 60 |
+  .startupProbe = {
+      "failureThreshold": 3,
+      "periodSeconds": 60,
+      "timeoutSeconds": 3,
+      "httpGet": {
+        "port": "healthz",
+        "path": "/healthz"}
+      }
+  )' -i "${ASSETS_DIR}/03-topolvm.yaml"
+
+  # Patch topolvm-node DaemonSet with probes
+  # echo 'Patching topolvm-node DaemonSet with longer startup delay to allow dns to start'
+  yq 'with(select(.kind == "DaemonSet" and .metadata.name == "topolvm-node").spec.template.spec.containers[] | select(.name == "topolvm-node");
+  .startupProbe = {
+      "failureThreshold": 60,
+      "periodSeconds": 2,
+      "timeoutSeconds": 3,
+      "httpGet": {
+        "port": "healthz",
+        "path": "/healthz"
+      }
+    })' -i "${ASSETS_DIR}/03-topolvm.yaml"
 
   # Generate kustomize
   cat >"${ASSETS_DIR}/kustomization.yaml" <<'EOF'
 apiVersion: kustomize.config.k8s.io/v1beta1
 kind: Kustomization
-resources:
-  - 01-namespace.yaml
-  - 02-cert-manager.yaml
-  - 03-topolvm.yaml
-EOF
-
-  cat >"${ASSETS_DIR}/kustomization.yaml" <<'EOF'
-apiVersion: kustomize.config.k8s.io/v1beta1
-kind: Kustomization
 resources:
   - 01-namespace.yaml
   - 02-cert-manager.yaml