Add ServiceMonitor support

Author: kempspo

charts/trino/README.md

@@ -640,6 +640,15 @@ Fast distributed SQL query engine for big data analytics that helps you explore
           value: '$2'
           help: 'ThreadCount (java.lang<type=Threading><>ThreadCount)'
           type: UNTYPED
+* `serviceMonitor.enabled` - bool, default: `false`  
+
+  Set to true to create resources for the [prometheus-operator](https://github.com/prometheus-operator/prometheus-operator).
+* `serviceMonitor.labels` - object, default: `{"prometheus":"kube-prometheus"}`  
+
+  Labels for serviceMonitor, so that Prometheus can select it
+* `serviceMonitor.interval` - string, default: `"30s"`  
+
+  The serviceMonitor web endpoint interval
 * `commonLabels` - object, default: `{}`  
 
   Labels that get applied to every resource's metadata

charts/trino/templates/servicemonitor.yaml

@@ -0,0 +1,22 @@
+{{- if .Values.serviceMonitor.enabled -}}
+apiVersion: monitoring.coreos.com/v1
+kind: ServiceMonitor
+metadata:
+  name: {{ template "trino.fullname" . }}
+  namespace: {{ .Release.Namespace }}
+  labels:
+    {{- include "trino.labels" . | nindent 4 }}
+    {{- if .Values.serviceMonitor.labels }}
+    {{- toYaml .Values.serviceMonitor.labels | nindent 4 }}
+    {{- end }}
+spec:
+  selector:
+    matchLabels:
+      {{- include "trino.selectorLabels" . | nindent 6 }}
+  namespaceSelector:
+    matchNames:
+      - {{ .Release.Namespace }}
+  endpoints:
+    - port: jmx-exporter
+      interval: {{ .Values.serviceMonitor.interval }}
+{{- end }}

charts/trino/templates/tests/test-connection.yaml

@@ -7,7 +7,7 @@ metadata:
     app.kubernetes.io/component: test
     test: connection
   annotations:
-    "helm.sh/hook": test-success
+    "helm.sh/hook": test
 spec:
   containers:
     - name: cli
@@ -25,8 +25,8 @@ spec:
       - --password
     {{- end }}
       - --debug
-      - --network-logging=BODY
       - --execute=SELECT 1
+      - --no-progress
     {{- if eq .Values.server.config.authenticationType "PASSWORD" }}
       env:
         - name: TRINO_PASSWORD

charts/trino/templates/tests/test-jmx.yaml

@@ -1,4 +1,4 @@
-{{- if .Values.jmx.exporter.enabled }}
+{{- if or .Values.jmx.exporter.enabled .Values.serviceMonitor.enabled }}
 apiVersion: v1
 kind: Pod
 metadata:
@@ -8,7 +8,7 @@ metadata:
     app.kubernetes.io/component: test
     test: jmx
   annotations:
-    "helm.sh/hook": test-success
+    "helm.sh/hook": test
 spec:
   containers:
     {{- if .Values.jmx.exporter.enabled }}
@@ -18,5 +18,50 @@ spec:
       args:
       - curl -s {{ include "trino.fullname" . }}.{{ .Release.Namespace }}:5556 | grep -q trino
     {{- end }}
+    {{- if .Values.serviceMonitor.enabled }}
+    - name: service-monitor
+      image: python:3-slim
+      command: ["python", "/tests/test.py"]
+      volumeMounts:
+        - name: tests
+          mountPath: /tests
+    {{- end }}
+  volumes:
+    - name: tests
+      configMap:
+        name: {{ include "trino.fullname" . }}-test-jmx
   restartPolicy: Never
+---
+apiVersion: v1
+kind: ConfigMap
+metadata:
+  name: {{ include "trino.fullname" . }}-test-jmx
+  labels:
+    {{- include "trino.labels" . | nindent 4 }}
+    app.kubernetes.io/component: test
+    test: jmx
+data:
+  test.py: |
+    from urllib.request import urlopen
+    from urllib.error import URLError, HTTPError
+    import json
+
+    url = "http://prometheus-operator-kube-p-prometheus:9090/api/v1/targets?scrapePool=serviceMonitor/{{ .Release.Namespace }}/{{ include "trino.fullname" . }}/0&state=active"
+    while True:
+      try:
+        with urlopen(url) as response:
+          data = json.load(response)
+      except (URLError, HTTPError) as e:
+        print("Error fetching targets, Prometheus service might not be ready: ", e)
+        continue
+
+      try:
+        service_name = data["data"]["activeTargets"][0]["discoveredLabels"]["__meta_kubernetes_service_name"]
+      except (KeyError, IndexError) as e:
+        print("Invalid Prometheus response: ", e)
+        continue
+
+      if service_name == "{{ include "trino.fullname" . }}":
+        print("Found expected service in Prometheus targets")
+        break
 {{- end }}

charts/trino/values.yaml

@@ -763,6 +763,15 @@ jmx:
     #         help: 'ThreadCount (java.lang<type=Threading><>ThreadCount)'
     #         type: UNTYPED
 
+serviceMonitor:
+  # serviceMonitor.enabled -- Set to true to create resources for the [prometheus-operator](https://github.com/prometheus-operator/prometheus-operator).
+  enabled: false
+  # serviceMonitor.labels -- Labels for serviceMonitor, so that Prometheus can select it
+  labels:
+    prometheus: kube-prometheus
+  # serviceMonitor.interval -- The serviceMonitor web endpoint interval
+  interval: "30s"
+
 # -- Labels that get applied to every resource's metadata
 commonLabels: {}
 

test-values.yaml

@@ -156,6 +156,12 @@ jmx:
         - pattern: 'trino.memory*'
         - pattern: 'trino.execution<name=QueryManager>*'
 
+serviceMonitor:
+  enabled: true
+  selector:
+    prometheus: default
+  interval: "30s"
+
 ingress:
   enabled: true
   hosts:

test.sh

@@ -19,7 +19,7 @@ function join_by {
 # default to randomly generated namespace, same as chart-testing would do, but we need to load secrets into the same namespace
 NAMESPACE=trino-$(LC_ALL=C tr -dc 'a-z0-9' </dev/urandom | head -c 6 || true)
 HELM_EXTRA_SET_ARGS=
-CT_ARGS=(--charts=charts/trino --skip-clean-up)
+CT_ARGS=(--charts=charts/trino --skip-clean-up --helm-extra-args="--timeout 2m")
 CLEANUP_NAMESPACE=true
 TEST_NAMES=(default single_node complete_values)
 
@@ -71,8 +71,18 @@ openssl req -new -newkey rsa:4096 -days 365 -nodes -x509 \
     -addext "subjectAltName=DNS:localhost,DNS:*.$NAMESPACE,DNS:*.$NAMESPACE.svc,DNS:*.$NAMESPACE.svc.cluster.local,IP:127.0.0.1" \
     -keyout cert.key -out cert.crt
 
-kubectl create namespace "$NAMESPACE"
-kubectl -n "$NAMESPACE" create secret tls certificates --cert=cert.crt --key=cert.key
+kubectl create namespace "$NAMESPACE" --dry-run=client --output yaml | kubectl apply --filename -
+kubectl -n "$NAMESPACE" create secret tls certificates --cert=cert.crt --key=cert.key --dry-run=client --output yaml | kubectl apply --filename -
+
+# only install the Prometheus Helm chart when running the `complete_values` test
+if printf '%s\0' "${TEST_NAMES[@]}" | grep -qwz complete_values; then
+    helm repo add prometheus-community https://prometheus-community.github.io/helm-charts
+    helm upgrade --install prometheus-operator prometheus-community/kube-prometheus-stack -n "$NAMESPACE" \
+        --version "60.0.2" \
+        --set prometheus.prometheusSpec.serviceMonitorSelectorNilUsesHelmValues=false \
+        --set grafana.enabled=false
+    kubectl rollout status --watch deployments -l release=prometheus-operator -n "$NAMESPACE"
+fi
 
 CT_ARGS+=(--namespace "$NAMESPACE")
 
@@ -84,21 +94,23 @@ for test_name in "${TEST_NAMES[@]}"; do
     if ! time ct install "${CT_ARGS[@]}" --helm-extra-set-args "$HELM_EXTRA_SET_ARGS ${testCases[$test_name]}"; then
         echo 1>&2 "❌ Test $test_name failed"
         echo 1>&2 "Test logs:"
-        kubectl --namespace "$NAMESPACE" logs --tail=-1 --selector app.kubernetes.io/component=test
+        kubectl --namespace "$NAMESPACE" logs --tail=-1 --selector app.kubernetes.io/component=test --all-containers=true
         result=1
     else
         echo 1>&2 "✅ Test $test_name completed"
     fi
     if [ "$CLEANUP_NAMESPACE" == "true" ]; then
-        for release in $(helm --namespace "$NAMESPACE" ls --all --short); do
+        for release in $(helm --namespace "$NAMESPACE" ls --all --short | grep -v 'prometheus-operator'); do
             echo 1>&2 "Cleaning up Helm release $release"
             helm --namespace "$NAMESPACE" delete "$release"
         done
     fi
 done
 
 if [ "$CLEANUP_NAMESPACE" == "true" ]; then
+    helm -n "$NAMESPACE" uninstall prometheus-operator --ignore-not-found
     kubectl delete namespace "$NAMESPACE"
+    kubectl delete crd $(kubectl api-resources --api-group=monitoring.coreos.com --output name)
 fi
 
 exit $result