feat: add database snapshot functionality to Helm chart

helm/charts/determined/templates/db-deployment.yaml

@@ -61,6 +61,11 @@ spec:
       volumes:
         - name: determined-db-storage
           persistentVolumeClaim:
-            claimName: determined-db-pvc-{{ .Release.Name }}
+            claimName: 
+              {{ if .Values.db.claimName }}
+              determined-db-pvc-{{ .Values.db.claimName }}
+              {{- else -}}
+              determined-db-pvc-{{ .Release.Name }}
+              {{ end }}
       {{ end }}
-{{ end}}
+{{ end }}

helm/charts/determined/templates/db-persistent-volume-claim.yaml

@@ -3,14 +3,25 @@
 apiVersion: v1
 kind: PersistentVolumeClaim
 metadata:
-  name: determined-db-pvc-{{ .Release.Name }}
+  name: 
+    {{ if and .Release.IsUpgrade .Values.db.restoreSnapshotName }}
+    determined-db-pvc-{{ .Values.db.restoreSnapshotName }}
+    {{- else -}}
+    determined-db-pvc-{{ .Release.Name }}
+    {{ end }}
   namespace: {{ .Release.Namespace }}
   labels:
     app: determined-db-{{ .Release.Name }}
     release: {{ .Release.Name }}
 spec:
   accessModes:
     - ReadWriteOnce
+  {{- if and .Release.IsUpgrade  .Values.db.restoreSnapshotName }}
+  dataSource:
+    name: determined-db-snapshot-{{ .Values.db.restoreSnapshotName }}
+    kind: VolumeSnapshot
+    apiGroup: snapshot.storage.k8s.io
+  {{ end }}
   resources:
     requests:
       storage: {{ required  "A valid Values.db.storageSize entry is required!" .Values.db.storageSize }}

helm/charts/determined/templates/db-volume-snapshot-class.yaml

@@ -0,0 +1,6 @@
+apiVersion: snapshot.storage.k8s.io/v1
+kind: VolumeSnapshotClass
+metadata:
+  name: determined-db-snapshot-class
+driver: pd.csi.storage.gke.io
+deletionPolicy: Retain

helm/charts/determined/templates/db-volume-snapshot.yaml

@@ -0,0 +1,19 @@
+{{- if and .Release.IsUpgrade .Values.db.snapshotName }}
+apiVersion: snapshot.storage.k8s.io/v1
+kind: VolumeSnapshot
+metadata:
+  name: determined-db-snapshot-{{ .Values.db.snapshotName }}
+  namespace: default
+  labels:
+    app: determined-db-{{ .Release.Name }}
+    release: {{ .Release.Name }}
+spec:
+  volumeSnapshotClassName: determined-db-snapshot-class
+  source:
+    persistentVolumeClaimName: 
+      {{ if .Values.db.claimName }}
+      determined-db-pvc-{{ .Values.db.claimName }}
+      {{- else -}}
+      determined-db-pvc-{{ .Release.Name }}
+      {{ end }}
+{{ end }}

helm/charts/determined/values.yaml

@@ -196,6 +196,23 @@ db:
   # resourceType: <secret/configMap>
   # certResourceName: <secret/configMap name>
 
+  # claimName refers to the persistent volume claim name for the database.
+  # If left undefined, a new PVC is created, named "determined-db-pvc-{{ .Release.name }}".
+  # Otherwise, use the pvc provided, given that it exists in the current context.
+  # claimName:
+
+  # snapshotName refers to the volume snapshot name, a backup of the database's persistent volume.
+  # If defined, Helm will "snapshot" the DB upon its next upgrade, creating a volumeSnapshot called
+  # `determined-db-snapsnot-{{ .Values.snapshotName}}`. This CANNOT be the name of a persistent volume claim
+  # that already exists, like the `determined-db-pvc-{{Release.Name}}`.
+  # snapshotName: determined1
+
+  # restoreSnapshotName refers to the volume snapshot name which you wish to restore the DB to.
+  # This only applies during a Helm upgrade, and if defined. This CANNOT be the name of a PVC that
+  # already exists, like the {{Release.Name}}. A new persistent volume & claim will be created, named
+  # `determined-db-pvc-{{ .Values.restoreSnapshotName}}`, restoring the data from 
+  # `determined-db-snapsnot-{{ .Values.restoreSnapshotName}}`.
+  # restoreSnapshotName:
 
 # checkpointStorage controls where checkpoints are stored. Supported types include `shared_fs`,
 # `gcs`, and `s3`.