From a6a78c9ef229277bbd451e61a23c12e3f77d2fec Mon Sep 17 00:00:00 2001
From: Ilya Kreymer <ikreymer@users.noreply.github.com>
Date: Mon, 13 Nov 2023 10:02:05 -0800
Subject: [PATCH] node affinity: set to required instead of preferred to keep
 crawlers on dedicated infrastructure (#1366)

Previously, the crawler pods use preferred node affinity, instead of
required node affinity. This results in crawler nodes running on the
main node pool. Instead, we want to ensure crawler nodes are running on
dedicated node pool (if configured).
- Converts 'preferred node affinity' to 'required node affinity' for
the node pool, while keeping preferred pod affinity for keeping all
crawler / redis pods together.
- For profiles, updates to same node affinity, and also adds
resource constraint to match a single crawler for profile browser,
which did not have resource constraints.
---
 chart/app-templates/crawler.yaml        | 20 ++++++++++++--------
 chart/app-templates/profilebrowser.yaml | 17 +++++++++++++----
 2 files changed, 25 insertions(+), 12 deletions(-)

diff --git a/chart/app-templates/crawler.yaml b/chart/app-templates/crawler.yaml
index a85eac89..67d8c588 100644
--- a/chart/app-templates/crawler.yaml
+++ b/chart/app-templates/crawler.yaml
@@ -60,24 +60,28 @@ spec:
         claimName: {{ name }}
 
   affinity:
+{% if crawler_node_type %}
     nodeAffinity:
-      preferredDuringSchedulingIgnoredDuringExecution:
-        - weight: 1
-          preference:
-            matchExpressions:
+      requiredDuringSchedulingIgnoredDuringExecution:
+        nodeSelectorTerms:
+          - matchExpressions:
             - key: nodeType
               operator: In
               values:
                 - "{{ crawler_node_type }}"
+{% endif %}
 
     podAffinity:
       preferredDuringSchedulingIgnoredDuringExecution:
-        - weight: 2
+        - weight: 10
           podAffinityTerm:
-            topologyKey: "failure-domain.beta.kubernetes.io/zone"
+            topologyKey: "kubernetes.io/hostname"
             labelSelector:
-              matchLabels:
-                crawl: {{ id }}
+              matchExpressions:
+              - key: crawl
+                operator: In
+                values:
+                - {{ id }}
 
   tolerations:
     - key: nodeType
diff --git a/chart/app-templates/profilebrowser.yaml b/chart/app-templates/profilebrowser.yaml
index 828beefa..335f705c 100644
--- a/chart/app-templates/profilebrowser.yaml
+++ b/chart/app-templates/profilebrowser.yaml
@@ -18,16 +18,17 @@ spec:
 
   restartPolicy: OnFailure
 
+{% if crawler_node_type %}
   affinity:
     nodeAffinity:
-      preferredDuringSchedulingIgnoredDuringExecution:
-        - weight: 1
-          preference:
-            matchExpressions:
+      requiredDuringSchedulingIgnoredDuringExecution:
+        nodeSelectorTerms:
+          - matchExpressions:
             - key: nodeType
               operator: In
               values:
                 - "{{ crawler_node_type }}"
+{% endif %}
 
   tolerations:
     - key: nodeType
@@ -74,3 +75,11 @@ spec:
         - name: CHROME_FLAGS
           value: "--proxy-server=socks5://{{ crawler_socks_proxy_host }}:{{ crawler_socks_proxy_port | default('9050') }}"
       {% endif %}
+
+      resources:
+        limits:
+          memory: "{{ crawler_memory }}"
+
+        requests:
+          cpu: "{{ crawler_cpu }}"
+          memory: "{{ crawler_memory }}"