feat: collecting metrics from controller

Author: 0x5457

charts/tensor-fusion/templates/controller-deployment.yaml

@@ -31,6 +31,10 @@ spec:
       containers:
         - name: controller
           image: "{{ .Values.controller.image.repository }}:{{ .Values.controller.image.tag | default .Chart.AppVersion }}"
+          command:
+            - /manager
+            - -metrics-bind-address
+            - :9000
           livenessProbe:
             {{- toYaml .Values.controller.livenessProbe | nindent 12 }}
           readinessProbe:
@@ -43,6 +47,23 @@ spec:
               mountPath: /tmp/k8s-webhook-server/serving-certs
             - name: config
               mountPath: /etc/tensor-fusion
+        - name: vector
+          image: docker.io/timberio/vector:nightly-2025-01-07-debian
+          env:
+            - name: NODE_NAME
+              valueFrom:
+                fieldRef:
+                  apiVersion: v1
+                  fieldPath: spec.nodeName
+          resources:
+            requests:
+              cpu: 50m
+              memory: 64Mi
+          volumeMounts:
+            - name: logs
+              mountPath: /logs
+            - name: vector-config
+              mountPath: /etc/vector
       volumes:
         - name: cert
           secret:
@@ -57,6 +78,12 @@ spec:
           configMap:
             name: {{ include "tensor-fusion.fullname" . }}-config
             defaultMode: 420
+        - name: vector-config
+          configMap:
+            name: {{ include "tensor-fusion.fullname" . }}-vector-config
+            defaultMode: 420
+        - name: logs
+          emptyDir: {}
       {{- with .Values.controller.affinity }}
       affinity:
         {{- toYaml . | nindent 8 }}

charts/tensor-fusion/templates/hypervisor-daemonset.yaml

@@ -54,11 +54,11 @@ spec:
         - name: vector
           image: docker.io/timberio/vector:nightly-2025-01-07-debian
           env:
-            - name: NODE_IP
+            - name: NODE_NAME
               valueFrom:
                 fieldRef:
                   apiVersion: v1
-                  fieldPath: status.hostIP
+                  fieldPath: spec.nodeName
           resources:
             requests:
               cpu: 50m

charts/tensor-fusion/templates/hypervisor-vector.yaml renamed to charts/tensor-fusion/templates/vector.yaml

@@ -11,8 +11,11 @@ data:
         data_dir: /logs
         include:
           - /logs/metrics.log
-      in:
-        type: host_metrics
+      
+      controller_metrics:
+        type: prometheus_scrape
+        endpoints:
+          - http://localhost:9000/metrics
 
     transforms:
       parse_influx:
@@ -28,7 +31,7 @@ data:
           - parse_influx
         source: |
           .namespace = "tf"
-          .tags.nodeip = "${NODE_IP}"
+          .tags.nodeName = "${NODE_NAME}"
       
       log_to_metric:
         type: log_to_metric
@@ -37,10 +40,15 @@ data:
         all_metrics: true
         metrics: []
     sinks:
-      
-      sink_greptimedb_metrics:
+      sink_greptimedb_hypervisor_metrics:
         type: greptimedb_metrics
         inputs:
           - log_to_metric
         new_naming: false
         endpoint: {{ .Values.hypervisor.greptimedbEendpoint }}
+
+      sink_greptimedb_controller_metrics:
+        type: prometheus_remote_write
+        inputs:
+          - controller_metrics
+        endpoint: http://{{ .Values.hypervisor.greptimedbEendpoint }}/v1/prometheus/write?db=public