Add service monitor in tei and tgi

Signed-off-by: Sanket Sudake <[email protected]>
infracloudio · Jul 19, 2024 · 8b76bb4 · 8b76bb4
1 parent 646dbac
commit 8b76bb4
Show file tree

Hide file tree

Showing 6 changed files with 86 additions and 2 deletions.
diff --git a/charts/text-embeddings-inference/Chart.yaml b/charts/text-embeddings-inference/Chart.yaml
@@ -15,7 +15,7 @@ type: application
 # This is the chart version. This version number should be incremented each time you make changes
 # to the chart and its templates, including the app version.
 # Versions are expected to follow Semantic Versioning (https://semver.org/)
-version: 0.1.4
+version: 0.1.5
 
 # This is the version number of the application being deployed. This version number should be
 # incremented each time you make changes to the application. Versions are not expected to

diff --git a/charts/text-embeddings-inference/templates/servicemonitor.yaml b/charts/text-embeddings-inference/templates/servicemonitor.yaml
@@ -0,0 +1,28 @@
+{{- if .Values.serviceMonitor.enabled -}}
+apiVersion: monitoring.coreos.com/v1
+kind: ServiceMonitor
+metadata:
+  name: {{ include "text-embeddings-inference.fullname" . }}-servicemonitor
+  {{- if .Values.serviceMonitor.namespace }}
+  namespace: {{ .Values.serviceMonitor.namespace }}
+  {{- end }}
+  labels:
+    {{- include "text-embeddings-inference.labels" . | nindent 4 }}
+    {{- with .Values.serviceMonitor.additionalLabels }}
+    {{- toYaml . | nindent 4 }}
+    {{- end }}
+spec:
+  selector:
+    matchLabels:
+      {{- include "text-embeddings-inference.selectorLabels" . | nindent 6 }}
+  endpoints:
+    - port: http
+      interval: {{ .Values.serviceMonitor.interval }}
+      scrapeTimeout: {{ .Values.serviceMonitor.scrapeTimeout }}
+      path: {{ .Values.serviceMonitor.path }}
+  jobLabel: {{ include "text-embeddings-inference.fullname" . }}
+  namespaceSelector:
+    matchNames:
+      - {{ .Release.Namespace }}
+{{- end -}}
+
diff --git a/charts/text-embeddings-inference/values.yaml b/charts/text-embeddings-inference/values.yaml
@@ -65,6 +65,20 @@ service:
   type: ClusterIP
   port: 80
 
+serviceMonitor:
+  # Enable the creation of a ServiceMonitor resource
+  enabled: false
+  # Specify the namespace the ServiceMonitor resource should be created in
+  namespace: ""
+  # Specify the interval at which metrics should be scraped
+  interval: 30s
+  # Specify the scrape timeout
+  scrapeTimeout: 10s
+  # path to scrape for metrics
+  path: /metrics
+  # additional labels to add to the ServiceMonitor
+  additionalLabels: {}
+
 ingress:
   enabled: false
   className: ""

diff --git a/charts/text-generation-inference/Chart.yaml b/charts/text-generation-inference/Chart.yaml
@@ -15,7 +15,7 @@ type: application
 # This is the chart version. This version number should be incremented each time you make changes
 # to the chart and its templates, including the app version.
 # Versions are expected to follow Semantic Versioning (https://semver.org/)
-version: 0.1.3
+version: 0.1.4
 
 # This is the version number of the application being deployed. This version number should be
 # incremented each time you make changes to the application. Versions are not expected to

diff --git a/charts/text-generation-inference/templates/servicemonitor.yaml b/charts/text-generation-inference/templates/servicemonitor.yaml
@@ -0,0 +1,28 @@
+{{- if .Values.serviceMonitor.enabled -}}
+apiVersion: monitoring.coreos.com/v1
+kind: ServiceMonitor
+metadata:
+  name: {{ include "text-generation-inference.fullname" . }}-servicemonitor
+  {{- if .Values.serviceMonitor.namespace }}
+  namespace: {{ .Values.serviceMonitor.namespace }}
+  {{- end }}
+  labels:
+    {{- include "text-generation-inference.labels" . | nindent 4 }}
+    {{- with .Values.serviceMonitor.additionalLabels }}
+    {{- toYaml . | nindent 4 }}
+    {{- end }}
+spec:
+  selector:
+    matchLabels:
+      {{- include "text-generation-inference.selectorLabels" . | nindent 6 }}
+  endpoints:
+    - port: http
+      interval: {{ .Values.serviceMonitor.interval }}
+      scrapeTimeout: {{ .Values.serviceMonitor.scrapeTimeout }}
+      path: {{ .Values.serviceMonitor.path }}
+  jobLabel: {{ include "text-generation-inference.fullname" . }}
+  namespaceSelector:
+    matchNames:
+      - {{ .Release.Namespace }}
+{{- end -}}
+
diff --git a/charts/text-generation-inference/values.yaml b/charts/text-generation-inference/values.yaml
@@ -65,6 +65,20 @@ service:
   type: ClusterIP
   port: 80
 
+serviceMonitor:
+  # Enable the creation of a ServiceMonitor resource
+  enabled: false
+  # Specify the namespace the ServiceMonitor resource should be created in
+  namespace: ""
+  # Specify the interval at which metrics should be scraped
+  interval: 30s
+  # Specify the scrape timeout
+  scrapeTimeout: 10s
+  # path to scrape for metrics
+  path: /metrics
+  # additional labels to add to the ServiceMonitor
+  additionalLabels: {}
+
 ingress:
   enabled: false
   className: ""