Merge branch 'main' into fix/mutex-msg

.github/workflows/ci-build.yaml

@@ -59,6 +59,9 @@ jobs:
               - Dockerfile
               - manifests/**
               - sdks/**
+              # example test suite
+              - examples/**
+              - hack/test-examples.sh
             codegen:
               - *common
               # generated files
@@ -196,47 +199,54 @@ jobs:
         include:
           - test: test-executor
             profile: minimal
-            useApi: false
+            use-api: false
           - test: test-corefunctional
             profile: minimal
-            useApi: false
+            use-api: false
           - test: test-functional
             profile: minimal
-            useApi: false
+            use-api: false
           - test: test-api
             profile: mysql
-            useApi: true
+            use-api: true
           - test: test-cli
             profile: mysql
-            useApi: true
+            use-api: true
           - test: test-cron
             profile: minimal
-            useApi: false
+            use-api: false
           - test: test-examples
             profile: minimal
-            useApi: false
+            use-api: false
           - test: test-plugins
             profile: plugins
-            useApi: false
+            use-api: false
           - test: test-java-sdk
             profile: minimal
-            useApi: true
+            use-api: true
           - test: test-python-sdk
             profile: minimal
-            useApi: true
+            use-api: true
           - test: test-executor
             install_k3s_version: v1.28.13+k3s1
             profile: minimal
-            useApi: false
+            use-api: false
           - test: test-corefunctional
             install_k3s_version: v1.28.13+k3s1
             profile: minimal
-            useApi: false
+            use-api: false
           - test: test-functional
             install_k3s_version: v1.28.13+k3s1
             profile: minimal
-            useApi: false
+            use-api: false
     steps:
+      - name: Free up disk space
+        run: |
+          printf "==> Available space before cleanup\n"
+          df -h
+          sudo rm -rf /usr/share/dotnet /usr/local/lib/android /opt/ghc /usr/local/.ghcup /opt/hostedtoolcache/CodeQL
+          printf "==> Available space after cleanup\n"
+          df -h
       - name: Install socat (needed by Kubernetes)
         run: sudo apt-get -y install socat
       - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # v4.1.1
@@ -264,7 +274,15 @@ jobs:
           else
             export INSTALL_K3S_VERSION=${{ matrix.install_k3s_version }}
           fi
-          curl -sfL https://get.k3s.io | INSTALL_K3S_CHANNEL=stable INSTALL_K3S_EXEC=--docker K3S_KUBECONFIG_MODE=644 sh -
+
+          cat >/tmp/kubelet.config <<EOT
+          apiVersion: kubelet.config.k8s.io/v1beta1
+          kind: KubeletConfiguration
+          imageMinimumGCAge: 1h
+          imageGCHighThresholdPercent: 100
+          EOT
+
+          curl -sfL https://get.k3s.io | INSTALL_K3S_CHANNEL=stable INSTALL_K3S_EXEC='--docker --kubelet-arg=config=/tmp/kubelet.config' K3S_KUBECONFIG_MODE=644 sh -
           until kubectl --kubeconfig=/etc/rancher/k3s/k3s.yaml cluster-info ; do sleep 10s ; done
           cp /etc/rancher/k3s/k3s.yaml /home/runner/.kubeconfig
           echo "- name: fake_token_user" >> $KUBECONFIG
@@ -294,22 +312,25 @@ jobs:
         run: make controller kit STATIC_FILES=false
       - name: Build CLI
         run: make cli STATIC_FILES=false
-        if: ${{matrix.useApi}}
+        if: ${{matrix.use-api}}
       - name: Start controller/API
         run: |
           make start PROFILE=${{matrix.profile}} \
             AUTH_MODE=client STATIC_FILES=false \
             LOG_LEVEL=info \
-            API=${{matrix.useApi}} \
+            API=${{matrix.use-api}} \
             UI=false \
             POD_STATUS_CAPTURE_FINALIZER=true > /tmp/argo.log 2>&1 &
       - name: Wait for controller to be up
-        run: make wait API=${{matrix.useApi}}
+        run: make wait PROFILE=${{matrix.profile}} API=${{matrix.use-api}}
         timeout-minutes: 5
       - name: Run tests ${{matrix.test}}
         run: make ${{matrix.test}} E2E_SUITE_TIMEOUT=20m STATIC_FILES=false
 
       # failure debugging below
+      - name: Failure debug - k3s logs
+        if: ${{ failure() }}
+        run: journalctl -u k3s
       - name: Failure debug - describe MinIO/MySQL deployment
         if: ${{ failure() }}
         run: |

Makefile

@@ -574,6 +574,10 @@ ifeq ($(API),true)
 	# Wait for Argo Server
 	until lsof -i :2746 > /dev/null ; do sleep 10s ; done
 endif
+ifeq ($(PROFILE),mysql)
+	# Wait for MySQL
+	until (: < /dev/tcp/localhost/3306) ; do sleep 10s ; done
+endif
 
 .PHONY: postgres-cli
 postgres-cli:

hack/test-examples.sh

@@ -13,12 +13,15 @@ done
 trap 'kubectl get wf' EXIT
 
 grep -lR 'workflows.argoproj.io/test' examples/* | while read f ; do
-  kubectl delete workflow -l workflows.argoproj.io/test
   echo "Running $f..."
-  kubectl create -f $f
-  name=$(kubectl get workflow -o name)
+  name=$(kubectl create -f $f -o name)
+
+  echo "Waiting for completion of $f..."
   kubectl wait --for=condition=Completed $name
   phase="$(kubectl get $name -o 'jsonpath={.status.phase}')"
   echo " -> $phase"
   test Succeeded == $phase
+
+  echo "Deleting $f..."
+  kubectl delete $name
 done

util/errors/errors.go

@@ -125,6 +125,9 @@ func isTransientNetworkErr(err error) bool {
 	} else if strings.Contains(errorString, "connect: connection refused") {
 		// If err is connection refused, retry.
 		return true
+	} else if strings.Contains(errorString, "invalid connection") {
+		// If err is invalid connection, retry.
+		return true
 	}
 
 	return false