Align default/staging memory config with production (7168 MiB heap, 8 Gi request) (#60309)

Align default/staging memory config with production to fix OOM crashes caused by Moda config layering applying the 4096 MiB heap limit instead of the production 7168 MiB limit.

Author: heiskr

config/kubernetes/default/deployments/webapp.yaml

@@ -24,7 +24,14 @@ spec:
           resources:
             requests:
               cpu: 1000m
-              memory: 4500Mi
+              # Absolute minimum to start app is 1000m
+              # Node is single-threaded but we want more CPUs
+              # for OS and image resizing, and other binary executions
+              # Better to increase replicas or memory than CPU
+              memory: 8.0Gi
+              # Absolute minimum to start app is 4500Mi
+              # Would increase with more pages, versions, or languages supported
+              # The additional memory helps during traffic surges
               # Keep ~600 MB headroom above --max-old-space-size in
               # config/moda/configuration/default/env.yaml for Node
               # off-heap memory and OS overhead.

config/moda/configuration/default/env.yaml

@@ -6,8 +6,8 @@ data:
   # V8 code cache, libuv) and OS overhead. If you change the memory
   # request in config/kubernetes/default/deployments/webapp.yaml,
   # update this value to match.
-  # Current: 4500 MiB request, 4096 MiB heap = 404 MiB headroom
-  NODE_OPTIONS: '--max-old-space-size=4096'
+  # Current: 8.0 Gi (8192 MiB) request, 7168 MiB heap = 1024 MiB headroom
+  NODE_OPTIONS: '--max-old-space-size=7168'
   PORT: '4000'
   ENABLED_LANGUAGES: 'en,es,ja,pt,zh,ru,fr,ko,de'
   RATE_LIMIT_MAX: '21'