[no-relnote] Update E2E test suite

Signed-off-by: Carlos Eduardo Arango Gutierrez <[email protected]>
Signed-off-by: Evan Lezar <[email protected]>
Signed-off-by: Carlos Eduardo Arango Gutierrez <[email protected]>

Author: ArangoGutierrez

.github/workflows/e2e.yaml

@@ -58,11 +58,14 @@ jobs:
     - name: Run e2e tests
       env:
         KUBECONFIG: ${{ github.workspace }}/kubeconfig
+        HELM_CHART: ${{ github.workspace }}/deployments/helm/nvidia-device-plugin
         E2E_IMAGE_REPO: ghcr.io/nvidia/k8s-device-plugin
         E2E_IMAGE_TAG: ${{ inputs.version }}
+        E2E_IMAGE_PULL_POLICY: Always
+        NVIDIA_DRIVER_ENABLED: true
         LOG_ARTIFACTS: ${{ github.workspace }}/e2e_logs
       run: |
-        make test-e2e
+        make -f tests/e2e/Makefile test
 
     - name: Archive test logs
       if: ${{ failure() }}
@@ -72,6 +75,13 @@ jobs:
         path: ./e2e_logs/
         retention-days: 15
 
+    - name: Archive Ginkgo logs
+      uses: actions/upload-artifact@v4
+      with:
+        name: ginkgo-logs
+        path: ginkgo.json
+        retention-days: 15
+
     - name: Send Slack alert notification
       id: slack
       if: false
@@ -80,8 +90,10 @@ jobs:
         SLACK_BOT_TOKEN: ${{ secrets.SLACK_BOT_TOKEN }}
         SUMMARY_URL: https://github.com/${{github.repository}}/actions/runs/${{github.run_id}}
       with:
-        channel-id: ${{ secrets.SLACK_CHANNEL_ID }}
-        slack-message: |
-          :x: On repository ${{ github.repository }} the Workflow *${{ github.workflow }}* has failed.
+        method: chat.postMessage
+        token: ${{ secrets.SLACK_BOT_TOKEN }}
+        payload: |
+          channel: ${{ secrets.SLACK_CHANNEL_ID }}
+          text: ":x: On repository ${{ github.repository }} the Workflow *${{ github.workflow }}* has failed.
 
-          Details: ${{ env.SUMMARY_URL }}
+          Details: ${{ env.SUMMARY_URL }}"

.gitignore

@@ -6,6 +6,7 @@
 deployments/helm/gpu-feature-discovery
 cmd/gpu-feature-discovery/gfd-test-loop
 e2e_logs
-
+bin
 *.out
 *.log
+ginkgo.json

testdata/job-1.yaml

@@ -0,0 +1,25 @@
+apiVersion: batch/v1
+kind: Job
+metadata:
+  name: j-e2e-1
+  labels:
+    app.nvidia.com: k8s-device-plugin-test-app
+spec:
+  template:
+    metadata:
+      name: gpu-pod
+    spec:
+      restartPolicy: Never
+      containers:
+        - name: cuda-container
+          image: nvcr.io/nvidia/k8s/cuda-sample:nbody-cuda11.7.1-ubuntu18.04
+          args:
+            - "--benchmark"
+            - "--numbodies=10000"
+          resources:
+            limits:
+              nvidia.com/gpu: "1"
+      tolerations:
+        - key: "nvidia.com/gpu"
+          operator: "Exists"
+          effect: "NoSchedule"

tests/e2e/Makefile

@@ -1,4 +1,4 @@
-# Copyright (c) 2023, NVIDIA CORPORATION.  All rights reserved.
+# Copyright (c) 2023, NVIDIA CORPORATION. All rights reserved.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
@@ -20,29 +20,33 @@ GO_TEST_TIMEOUT ?= 30m
 
 include $(CURDIR)/versions.mk
 
-DRIVER_ENABLED ?= true
+NVIDIA_DRIVER_ENABLED ?= true
 
 E2E_IMAGE_REPO ?= $(REGISTRY)/$(DRIVER_NAME)
 E2E_IMAGE_TAG ?= $(VERSION)
 E2E_IMAGE_PULL_POLICY ?= IfNotPresent
 HELM_CHART ?= $(CURDIR)/deployments/helm/nvidia-device-plugin
 LOG_ARTIFACTS ?= $(CURDIR)/e2e_logs
 
-.PHONY: test
-test:
-	@if [ -z ${KUBECONFIG} ]; then \
-		echo "[ERR] KUBECONFIG missing, must be defined"; \
-		exit 1; \
-	fi
-	cd $(CURDIR)/tests/e2e && $(GO_CMD) test -timeout $(GO_TEST_TIMEOUT) -v . -args \
-		-kubeconfig=$(KUBECONFIG) \
-		-driver-enabled=$(DRIVER_ENABLED) \
-		-image.repo=$(E2E_IMAGE_REPO) \
-		-image.tag=$(E2E_IMAGE_TAG) \
-		-image.pull-policy=$(E2E_IMAGE_PULL_POLICY) \
-		-log-artifacts=$(LOG_ARTIFACTS) \
-		-helm-chart=$(HELM_CHART) \
-		-helm-log-file=$(LOG_ARTIFACTS)/helm.log \
-		-ginkgo.focus="\[nvidia\]" \
-		-test.timeout=1h \
-		-ginkgo.v
+# Test configuration
+GINKGO_COMMON_ARGS := -v --fail-on-pending --randomize-all --trace
+GINKGO_REPORT_ARGS := --json-report=$(LOG_ARTIFACTS)/report.json --junit-report=$(LOG_ARTIFACTS)/junit.xml
+
+.PHONY: ginkgo test clean-artifacts
+
+GINKGO_VERSION = $(shell grep -Eo "github.com/onsi/ginkgo/v2.*$$" ./tests/go.mod | sed -e 's&github.com/onsi/ginkgo/v2[[:space:]]&&g')
+ginkgo:
+	mkdir -p $(CURDIR)/bin
+	GOBIN=$(CURDIR)/bin go install github.com/onsi/ginkgo/v2/ginkgo@$(GINKGO_VERSION)
+
+# Create artifacts directory
+$(LOG_ARTIFACTS):
+	mkdir -p $(LOG_ARTIFACTS)
+
+# Clean artifacts
+clean-artifacts:
+	rm -rf $(LOG_ARTIFACTS)
+
+# Run tests
+test: ginkgo $(LOG_ARTIFACTS)
+	$(CURDIR)/bin/ginkgo $(GINKGO_COMMON_ARGS) $(GINKGO_REPORT_ARGS) $(GINKGO_ARGS) ./tests/e2e/...