fix(kontinuous): deploy Valkey via a direct Helm dependency

Drop the .kontinuous/charts/valkey wrapper subchart and alias hack.
Declare the upstream valkey-io/valkey-helm chart directly as a Helm
dependency in .kontinuous/Chart.yaml, which is the canonical kontinuous
pattern for external charts.

Values now pass straight through to the upstream chart — no 'server:'
nesting, labels read 'app.kubernetes.io/name: valkey' (not 'server').
Fixes the connection timeouts observed in the app pod: the previous
wrapper produced a subchart whose service/pod labels were shifted by
the alias, which could leave endpoints out of sync with the service
selector depending on kontinuous's build order.

Also drop stale 'Redis-compatible' wording from docs/comments.

Author: maxgfr

.kontinuous/Chart.yaml

@@ -0,0 +1,4 @@
+dependencies:
+  - name: valkey
+    repository: https://valkey.io/valkey-helm
+    version: 0.9.4

.kontinuous/charts/valkey/Chart.yaml

@@ -18,19 +18,18 @@ app:
       memory: 2G
 
 valkey:
-  server:
-    replica:
-      enabled: true
-      replicas: 2
-      persistence:
-        size: 1Gi
-    podDisruptionBudget:
-      enabled: true
-      maxUnavailable: 1
-    resources:
-      requests:
-        cpu: 100m
-        memory: 256Mi
-      limits:
-        cpu: 1
-        memory: 1Gi
+  replica:
+    enabled: true
+    replicas: 2
+    persistence:
+      size: 1Gi
+  podDisruptionBudget:
+    enabled: true
+    maxUnavailable: 1
+  resources:
+    requests:
+      cpu: 100m
+      memory: 256Mi
+    limits:
+      cpu: 1
+      memory: 1Gi

.kontinuous/charts/valkey/values.yaml

@@ -88,11 +88,30 @@ app:
 pg:
   ~chart: pg
 
-# Valkey (Redis-compatible) cache.
-# Deployed via the project-local `valkey` wrapper chart (.kontinuous/charts/valkey)
-# which pulls the official valkey-io/valkey-helm chart as a dependency.
-# Base defaults live in the wrapper's values.yaml; per-env overrides below.
-valkey: {}
+# Valkey cache. Official valkey-io/valkey-helm chart pulled as a Helm dependency
+# from .kontinuous/Chart.yaml. Values below flow straight through to that chart.
+valkey:
+  # Pin service name so VALKEY_URL (redis://valkey:6379) resolves in every namespace.
+  fullnameOverride: valkey
+  image:
+    tag: "9.0.2"
+  service:
+    type: ClusterIP
+    port: 6379
+  # Standalone by default. Prod overrides to master/replica topology.
+  replica:
+    enabled: false
+  dataStorage:
+    enabled: false
+  auth:
+    enabled: false
+  resources:
+    requests:
+      cpu: 50m
+      memory: 128Mi
+    limits:
+      cpu: 500m
+      memory: 512Mi
 
 clamav:
   ~chart: clamav

.kontinuous/env/prod/values.yaml

@@ -28,5 +28,5 @@ MAIL_ENABLED="true"
 SMTP_HOST="localhost"
 SMTP_PORT="1025"
 MAIL_FROM="no-reply@egapro.local"
-# Valkey (Redis-compatible) cache — optional, app works without it (in-memory fallback)
+# Valkey cache — optional, app works without it (in-memory fallback)
 VALKEY_URL="redis://localhost:6379"

.kontinuous/values.yaml

@@ -1,6 +1,6 @@
 // @ts-check
 /**
- * Custom Next.js cache handler backed by Valkey (Redis-compatible).
+ * Custom Next.js cache handler backed by Valkey.
  *
  * Implements the legacy cacheHandler interface (get, set, revalidateTag, resetRequestCache).
  * Used for ISR pages, fetch cache, route handlers, and image optimization.

packages/app/.env.example

@@ -89,7 +89,7 @@ export const env = createEnv({
 		SMTP_USER: z.string().optional(),
 		SMTP_PASS: z.string().optional(),
 		MAIL_FROM: z.string().default("no-reply@egapro.local"),
-		// Valkey (Redis-compatible) cache URL — optional. When absent, the custom
+		// Valkey cache URL — optional. When absent, the custom
 		// cache handler (cache-handler.cjs) gracefully degrades to no-op.
 		// The handler reads process.env.VALKEY_URL directly (it runs outside
 		// the app module graph), but we declare it here for validation and docs.