Merge pull request #521 from paulfantom/datasource-fix

{chart,scripts}: move datasource test to helm chart

.github/workflows/tests.yml

@@ -19,13 +19,13 @@ jobs:
         uses: helm/[email protected]
 
       - name: Run chart-testing (lint)
-        run: ct lint --config ct.yaml
+        run: make lint
 
   test:
     runs-on: ubuntu-latest
     strategy:
       matrix:
-        kube-version:
+        kube:
           - "1.23"
           - "1.24"
     steps:
@@ -39,23 +39,15 @@ jobs:
 
       - name: Start kuberenetes cluster with cert-manager
         env:
-          KUBE_VERSION: ${{ matrix.kube-version }}
+          KUBE_VERSION: ${{ matrix.kube }}
         run: |
           make cert-manager
 
       - name: Wait for cluster to finish bootstraping
         run: kubectl wait --for=condition=Ready pods --all --all-namespaces --timeout=300s
 
-      - name: Create namespace for installing the chart
-        run: kubectl create namespace tobs-test
-
-      - name: Run chart-testing (install)
-        run: ct install --config ct.yaml --namespace tobs-test
-
-      # TODO(onprem): Figure out a workaround to test datasources as ct just removes
-      # the chart release after a successful install.
-      # - name: Check datasources
-      #   run: make check-datasources
+      - name: Run e2e chart-testing
+        run: make e2e
 
   test-result:
     name: End-to-End Test Results

.gitignore

@@ -1,6 +1,6 @@
 .vscode/
 .idea/
 chart/Chart.lock
-chart/charts/*
+chart/charts
 tmp/
 manifests.yaml

Makefile

@@ -46,8 +46,14 @@ helm-install: cert-manager load-images  ## This is a phony target that is used t
 	helm dep up chart/
 	helm upgrade --install --wait --timeout 15m test chart/
 
-.PHONY: check-datasources
-	./scripts/check-datasources.sh
+.PHONY: lint
+lint:  ## Lint helm chart using ct (chart-testing).
+	ct lint --config ct.yaml
+
+.PHONY: e2e
+e2e:  ## Run e2e installation tests using ct (chart-testing).
+	kubectl create ns tobs-test
+	ct install --config ct.yaml --namespace tobs-test
 
 manifests.yaml:
 	helm template --namespace test test chart/ > $@

chart/ci/default-values.yaml

@@ -0,0 +1,35 @@
+# This is the default values for the chart to run in CI system and it can be used as
+# a base template for other *-values.yaml files. Only resource requests are nullified
+# to allow starting the stack on github action runner and prevent issues related
+# to pod scheduling due to insufficient host resources.
+
+timescaledb-single:
+  env:
+  - name: "TIMESCALEDB_TELEMETRY"
+    value: "off"
+  resources: null
+
+promscale:
+  resources: null
+
+kube-prometheus-stack:
+  alertmanager:
+    alertmanagerSpec:
+      resources: null
+  prometheusOperator:
+    prometheusConfigReloader:
+      resources: null
+    resources: null
+  prometheus:
+    prometheusSpec:
+      resources: null
+  grafana:
+    resources: null
+  kube-state-metrics:
+    resources: null
+  prometheus-node-exporter:
+    resources: null
+
+opentelemetry-operator:
+  manager:
+    resources: null