Revert "[SSR-1] Removal of Rendertron (#474)"

This reverts commit 7b58985.
TUM-Dev · Apr 1, 2023 · 66feef2 · 66feef2
1 parent b8cb11d
commit 66feef2
Show file tree

Hide file tree

Showing 5 changed files with 133 additions and 0 deletions.
diff --git a/deployment/k3s/templates/deployments/web/rendertron-deployment.yaml b/deployment/k3s/templates/deployments/web/rendertron-deployment.yaml
@@ -0,0 +1,68 @@
+apiVersion: apps/v1
+kind: Deployment
+metadata:
+  name: rendertron
+  labels:
+    app: navigatum
+    deployment: rendertron
+  namespace: {{ $.Values.namespace }}
+spec:
+  replicas: {{ if eq "nav.tum.de" $.Values.url }}2{{ else }}1{{ end }}
+  revisionHistoryLimit: 1
+  selector:
+    matchLabels:
+      app: navigatum
+      deployment: rendertron
+  strategy:
+    rollingUpdate:
+      maxSurge: {{ if eq "nav.tum.de" $.Values.url }}50%{{ else }}100%{{ end }}
+      maxUnavailable: {{ if eq "nav.tum.de" $.Values.url }}50%{{ else }}0%{{ end }}
+    type: RollingUpdate
+  template:
+    metadata:
+      labels:
+        app: navigatum
+        deployment: rendertron
+    spec:
+      affinity:
+        nodeAffinity:
+          requiredDuringSchedulingIgnoredDuringExecution:
+            nodeSelectorTerms:
+            - matchExpressions:
+              - key: tileserver
+                operator: In
+                values:
+                - ready
+      volumes:
+        - name: tmp-rendertron
+          emptyDir: {}
+      containers:
+        - name: rendertron
+          image: dockette/rendertron
+          imagePullPolicy: Always
+          ports:
+            - containerPort: 3000
+              name: rendertron
+          volumeMounts:
+            - mountPath: /tmp
+              name: tmp-rendertron
+          securityContext:
+            readOnlyRootFilesystem: true
+          resources:
+            requests:
+              cpu: 50m
+              memory: 200Mi
+            limits:
+              memory: 500Mi
+          livenessProbe:
+            httpGet:
+              path: /_ah/health
+              port: rendertron
+            failureThreshold: 1
+            periodSeconds: 60
+          startupProbe:
+            httpGet:
+              path: /_ah/health
+              port: rendertron
+            failureThreshold: 20
+            periodSeconds: 5
diff --git a/deployment/k3s/templates/networking/ingress.yaml b/deployment/k3s/templates/networking/ingress.yaml
@@ -37,6 +37,11 @@ spec:
       services:
         - name: cdn-svc
           port: 8000
+    - kind: Rule
+      match: Host(`{{ $.Values.url }}`) && PathPrefix(`/rendertron/`)
+      services:
+        - name: rendertron-svc
+          port: 9100
       middlewares:
         - name: stripprefix
     - kind: Rule

diff --git a/deployment/k3s/templates/networking/middleware/stripprefix-middleware.yaml b/deployment/k3s/templates/networking/middleware/stripprefix-middleware.yaml
@@ -8,4 +8,5 @@ metadata:
 spec:
   stripPrefix:
     prefixes:
+      - /rendertron
       - /maps
diff --git a/deployment/k3s/templates/networking/service.yaml b/deployment/k3s/templates/networking/service.yaml
@@ -108,3 +108,21 @@ spec:
     - name: webclient
       port: 9000
       targetPort: 80
+---
+apiVersion: v1
+kind: Service
+metadata:
+  name: rendertron-svc
+  labels:
+    app: navigatum
+    deployment: rendertron
+  namespace: {{ $.Values.namespace }}
+spec:
+  type: ClusterIP
+  selector:
+    app: navigatum
+    deployment: rendertron
+  ports:
+    - name: rendertron
+      port: 9100
+      targetPort: 3000
diff --git a/webclient/nginx.conf b/webclient/nginx.conf
@@ -44,6 +44,31 @@ http {
     "de" "de";
   }
 
+  # bot without js-execution workaround
+  # see https://github.com/GoogleChrome/rendertron/blob/main/middleware/src/middleware.ts#L24-L41
+  map $http_user_agent $no_js_exec {
+    default 0;
+    '~*Baiduspider' 1;
+    '~*bingbot' 1;
+    '~*Embedly' 1;
+    '~*facebookexternalhit' 1;
+    '~*LinkedInBot' 1;
+    '~*outbrain' 1;
+    '~*pinterest' 1;
+    '~*quora link preview' 1;
+    '~*rogerbot' 1;
+    '~*showyoubot' 1;
+    '~*Slackbot' 1;
+    '~*TelegramBot' 1;
+    '~*Twitterbot' 1;
+    '~*vkShare' 1;
+    '~*W3C_Validator' 1;
+    '~*WhatsApp' 1;
+    '~*Discordbot' 1;
+    '~*googlebot' 1;
+    '~*SemrushBot' 1;
+  }
+
   server {
     # default_server makes nginx not care about HOST, we hande this upstream by Traefik
     listen 80 default_server;
@@ -72,6 +97,22 @@ http {
       return 200 'healthy';
     }
 
+    if ($no_js_exec = 1) {
+      rewrite ^/(((api|search|about|view|campus|site|building|room)/?.*)?)$ /rendertron/$1;
+    }
+    # equivalent to:
+    #if ($uri ~ ^/((\.well-known|assets)/.*|(robots.txt|googlebef9161f1176c5e0.html|adds.txt|app-ads.txt|favicon.ico|health|404.html|50x.html))$){
+    #  set $no_js_exec 0;
+    #}
+
+    #render all pages, that could execute js via rendertron
+    location ^~ /rendertron/ {
+        proxy_set_header X-Real-IP  $remote_addr;
+        proxy_set_header X-Forwarded-For $remote_addr;
+        proxy_set_header User-Agent 'Rendertron';
+        proxy_pass https://rendertron.navigatum-scv.svc.cluster.local/rendertron/render/https://nav.tum.de$request_uri;
+    }
+
     location / {
       add_header Cache-Control no-cache; # disable caching, as we do not want to have theme/language related issues
       expires 360s; # 360s=5min