Add WAL failover custom path support for helm chart migration

Author: shreyaskm623

build/templates/cockroachdb-parent/charts/cockroachdb/values.yaml

@@ -265,6 +265,11 @@ cockroachdb:
           #
     # walFailoverSpec captures the configuration for WAL Failover.
     walFailoverSpec: {}
+      # name determines the prefix of the pvc name that we want to provide
+      # for the wal failover. This field was introduced during the migration process
+      # from the statefulset to the operator.
+      # name: failover-dir
+      #
       # status determines the possible values to WAL Failover configuration.
       # It has 3 possible values: "", "enable" and "disable".
       # status: ""
@@ -276,6 +281,14 @@ cockroachdb:
       # If not set, the default provisioner will be chosen (gp2 on AWS, standard on GKE).
       # storageClassName: ""
       #
+      # path is the absolute mount path inside the CockroachDB container to use for WAL failover side disk.
+      # Examples: 
+      #   - /cockroach/cockroach-failover
+      #   - /cockroach/custom/path
+      # IMPORTANT: Set it only for fresh deployments. Changing it on upgrades might cause issues with the deployment.
+      #
+      # path: "/cockroach/cockroach-failover"
+      #
     # podLabels captures additional labels to apply to CockroachDB pods.
     podLabels:
       app.kubernetes.io/component: cockroachdb

build/templates/cockroachdb-parent/charts/operator/values.yaml

@@ -9,7 +9,7 @@ image:
   # pullPolicy specifies the image pull policy.
   pullPolicy: IfNotPresent
   # tag is the image tag.
-  tag: "6f62639e9fee99d99b0387a9dccda84daa1a489b592b008f2f354ec57eae09ac"
+  tag: "24cc995b44f2ff4a2c30b8290d229602c8b5ad90c74d0843c246f701e30db872"
 # certificate defines the certificate settings for the Operator.
 certificate:
   # validForDays specifies the number of days the certificate is valid for.

cockroachdb-parent/charts/cockroachdb/values.yaml

@@ -266,6 +266,11 @@ cockroachdb:
           #
     # walFailoverSpec captures the configuration for WAL Failover.
     walFailoverSpec: {}
+      # name determines the prefix of the pvc name that we want to provide
+      # for the wal failover. This field was introduced during the migration process
+      # from the statefulset to the operator.
+      # name: failover-dir
+      #
       # status determines the possible values to WAL Failover configuration.
       # It has 3 possible values: "", "enable" and "disable".
       # status: ""
@@ -277,6 +282,14 @@ cockroachdb:
       # If not set, the default provisioner will be chosen (gp2 on AWS, standard on GKE).
       # storageClassName: ""
       #
+      # path is the absolute mount path inside the CockroachDB container to use for WAL failover side disk.
+      # Examples: 
+      #   - /cockroach/cockroach-failover
+      #   - /cockroach/custom/path
+      # IMPORTANT: Set it only for fresh deployments. Changing it on upgrades might cause issues with the deployment.
+      #
+      # path: "/cockroach/cockroach-failover"
+      #
     # podLabels captures additional labels to apply to CockroachDB pods.
     podLabels:
       app.kubernetes.io/component: cockroachdb

cockroachdb-parent/charts/operator/templates/operator.yaml

@@ -254,6 +254,7 @@ rules:
       - admissionregistration.k8s.io
     resources:
       - validatingwebhookconfigurations
+      - mutatingwebhookconfigurations
     verbs:
       - create
   - apiGroups:
@@ -265,6 +266,15 @@ rules:
     verbs:
       - get
       - patch
+  - apiGroups:
+      - admissionregistration.k8s.io
+    resources:
+      - mutatingwebhookconfigurations
+    resourceNames:
+      - cockroach-mutating-webhook-config
+    verbs:
+      - get
+      - patch
   # The "create" verb cannot be qualified with resourceNames, so grant the
   # unqualified permission so that the operator can create new CRDs. But only
   # allow the operator to get and patch its own CRDs.

cockroachdb-parent/charts/operator/values.yaml

@@ -10,7 +10,7 @@ image:
   # pullPolicy specifies the image pull policy.
   pullPolicy: IfNotPresent
   # tag is the image tag.
-  tag: "6f62639e9fee99d99b0387a9dccda84daa1a489b592b008f2f354ec57eae09ac"
+  tag: "24cc995b44f2ff4a2c30b8290d229602c8b5ad90c74d0843c246f701e30db872"
 # certificate defines the certificate settings for the Operator.
 certificate:
   # validForDays specifies the number of days the certificate is valid for.

pkg/migrate/build-manifests.go

@@ -172,6 +172,9 @@ func (m *Manifest) FromHelmChart() error {
 	}
 
 	input.pcrSpec = detectPCRFromInitJob(m.clientset, sts.Name, m.namespace)
+
+	// Cluster level WalSpec
+	var clusterWalSpec *v1alpha1.CrdbWalFailoverSpec
 	for nodeIdx := int32(0); nodeIdx < *sts.Spec.Replicas; nodeIdx++ {
 		podName := fmt.Sprintf("%s-%d", sts.Name, nodeIdx)
 		pod, err := m.clientset.CoreV1().Pods(m.namespace).Get(ctx, podName, metav1.GetOptions{})
@@ -183,6 +186,11 @@ func (m *Manifest) FromHelmChart() error {
 			return errors.Newf("pod %s isn't scheduled to a node", podName)
 		}
 
+		// Build the walFailoverSpec if applicable
+		buildWalFailoverSpec(ctx, m.clientset, sts, pod.Spec.NodeName, nodeIdx, &input)
+		if input.walFailoverSpec != nil && clusterWalSpec == nil {
+			clusterWalSpec = input.walFailoverSpec
+		}
 		nodeSpec := buildNodeSpecFromHelm(sts, pod.Spec.NodeName, input)
 		crdbNode := v1alpha1.CrdbNode{
 			TypeMeta: metav1.TypeMeta{