Add nodeAffinity to all logging components, add resources to fluent-bit, tuning

charts/kubezero-logging/Chart.yaml

@@ -2,7 +2,7 @@ apiVersion: v2
 name: kubezero-logging
 description: KubeZero Umbrella Chart for complete EFK stack
 type: application
-version: 0.6.1
+version: 0.6.2
 appVersion: 1.4.1
 home: https://kubezero.com
 icon: https://cdn.zero-downtime.net/assets/kubezero/logo-small-64.png

charts/kubezero-logging/templates/eck/elasticsearch.yaml

@@ -18,7 +18,7 @@ spec:
       {{- end }}
       {{- if .zone }}
       node.attr.zone: {{ .zone }}
-      cluster.routing.allocation.awareness.attributes: zone
+      cluster.routing.allocation.awareness.attributes: k8s_node_name,zone
       {{- end }}
       transport.compress: true
       {{- if .processors }}
@@ -66,8 +66,9 @@ spec:
                 matchLabels:
                   elasticsearch.k8s.elastic.co/cluster-name: {{ template "kubezero-lib.fullname" $ }}
               topologyKey: kubernetes.io/hostname
-          {{- if .zone }}
+          {{- if or .zone .nodeAffinity }}
           nodeAffinity:
+            {{- if .zone }}
             requiredDuringSchedulingIgnoredDuringExecution:
               nodeSelectorTerms:
               - matchExpressions:
@@ -75,6 +76,17 @@ spec:
                   operator: In
                   values:
                   - {{ .zone }}
+            {{- end }}
+            {{- if .nodeAffinity }}
+            preferredDuringSchedulingIgnoredDuringExecution:
+              - weight: 1
+                preference:
+                  matchExpressions:
+                  - key: {{ .nodeAffinity.key }}
+                    operator: In
+                    values:
+                    - {{ .nodeAffinity.value }}
+            {{- end }}
           {{- end }}
     count: {{ .count }}
     volumeClaimTemplates:

charts/kubezero-logging/templates/eck/kibana.yaml

@@ -29,6 +29,19 @@ spec:
             cpu: 200m
           limits:
             memory: 2Gi
+      {{- if .Values.kibana.nodeAffinity }}
+      affinity:
+        nodeAffinity:
+          preferredDuringSchedulingIgnoredDuringExecution:
+            - weight: 1
+              preference:
+                matchExpressions:
+                - key: {{ .Values.kibana.nodeAffinity.key }}
+                  operator: In
+                  values:
+                  - {{ .Values.kibana.nodeAffinity.value }}
+      {{- end }}
+
   http:
     tls:
       selfSignedCertificate:

charts/kubezero-logging/values-all.yaml

@@ -18,8 +18,9 @@ es:
       size: 512Gi
       class: ebs-sc-gp2-xfs
     zone: us-west-2a
-    jvm_heap: 4
-    cpu_limit: 2
+    nodeAffinity:
+      key: kubezero
+      value: logging
   s3Snapshot:
     enabled: true
     iamrole: "dfsf" # INSERT_CLOUDFORMATION_OUTPUT_ElasticSearchSnapshots
@@ -27,6 +28,9 @@ es:
   prometheus: true
 
 kibana: 
+  nodeAffinity:
+    key: kubezero
+    value: logging
   istio:
     enabled: true
     url: kibana.example.com

charts/kubezero-logging/values-fluentd.yaml

@@ -4,6 +4,17 @@ fluentd:
   metrics:
     enabled: true
 
+  affinity:
+    nodeAffinity:
+      preferredDuringSchedulingIgnoredDuringExecution:
+        - weight: 1
+          preference:
+            matchExpressions:
+            - key: kubezero
+              operator: In
+              values:
+              - logging
+
   url: fluentd.example.com
   istio:
     enabled: true

charts/kubezero-logging/values.yaml

@@ -22,6 +22,9 @@ es:
   #    size: 16Gi
   #    class: local-sc-xfs
   #  zone: us-west-2a
+  #  nodeAffinity:
+  #    key: node label name
+  #    value: node label value
   s3Snapshot:
     enabled: false
     iamrole: "" # INSERT_CLOUDFORMATION_OUTPUT_ElasticSearchSnapshots
@@ -31,6 +34,9 @@ es:
 kibana:
   count: 1
   #servicename: kibana.example.com
+  #nodeAffinity:
+  #  key: node label name
+  #  value: node label value
   istio:
     enabled: false
     gateway: "istio-system/ingressgateway"
@@ -86,10 +92,10 @@ fluentd:
 
   resources:
     requests:
-      cpu: 50m
+      cpu: 200m
       memory: 256Mi
     limits:
-      memory: 1024Mi
+      memory: 512Mi
 
   output:
     # Default should be "logging-kubezero-logging-es-http" if fullnameOverride is NOT used
@@ -184,8 +190,8 @@ fluentd:
             chunk_limit_size 8MB
             total_limit_size 4GB
             flush_mode interval
-            flush_thread_count 4
-            flush_interval 5s
+            flush_thread_count 2
+            flush_interval 10s
             flush_at_shutdown true
             retry_type exponential_backoff
             retry_timeout 300m
@@ -207,6 +213,13 @@ fluent-bit:
   - key: node-role.kubernetes.io/master
     effect: NoSchedule
 
+  resources:
+    requests:
+      cpu: 20m
+      memory: 16Mi
+    limits:
+      memory: 64Mi
+
   config:
     output:
       host: logging-fluentd