Switch agent from StatefulSet to Deployment (SURE-9419)

.github/scripts/deploy-fleet.sh

@@ -74,7 +74,7 @@ eventually helm upgrade --install fleet charts/fleet \
 # wait for controller and agent rollout
 kubectl -n cattle-fleet-system rollout status deploy/fleet-controller
 { grep -E -q -m 1 "fleet-agent-local.*1/1"; kill $!; } < <(kubectl get bundles -n fleet-local -w)
-kubectl -n cattle-fleet-system rollout status statefulset/fleet-agent
+kubectl -n cattle-fleet-system rollout status deployment/fleet-agent
 
 # label local cluster
 kubectl patch clusters.fleet.cattle.io -n fleet-local local --type=json -p '[{"op": "add", "path": "/metadata/labels/management.cattle.io~1cluster-display-name", "value": "local" }]'

.github/scripts/upgrade-rancher-fleet-to-dev-fleet.sh

@@ -46,6 +46,6 @@ sleep 5
 { grep -E -q -m 1 "fleet-agent-c.*1/1"; kill $!; } < <(kubectl get bundles -n fleet-default -w)
 
 kubectl config use-context k3d-downstream
-kubectl -n cattle-fleet-system rollout status statefulset/fleet-agent
+kubectl -n cattle-fleet-system rollout status deployment/fleet-agent
 
 helm list -A

charts/fleet-agent/templates/deployment.yaml

@@ -1,9 +1,8 @@
 apiVersion: apps/v1
-kind: StatefulSet
+kind: Deployment
 metadata:
   name: fleet-agent
 spec:
-  serviceName: fleet-agent
   selector:
     matchLabels:
       app: fleet-agent
@@ -18,6 +17,18 @@ spec:
           valueFrom:
             fieldRef:
               fieldPath: metadata.namespace
+        {{- if $.Values.leaderElection.leaseDuration }}
+        - name: CATTLE_ELECTION_LEASE_DURATION
+          value: {{$.Values.leaderElection.leaseDuration}}
+        {{- end }}
+        {{- if $.Values.leaderElection.retryPeriod }}
+        - name: CATTLE_ELECTION_RETRY_PERIOD
+          value: {{$.Values.leaderElection.retryPeriod}}
+        {{- end }}
+        {{- if $.Values.leaderElection.renewDeadline }}
+        - name: CATTLE_ELECTION_RENEW_DEADLINE
+          value: {{$.Values.leaderElection.renewDeadline}}
+        {{- end }}
         image: '{{ template "system_default_registry" . }}{{.Values.image.repository}}:{{.Values.image.tag}}'
         name: fleet-agent-register
         command:
@@ -51,6 +62,18 @@ spec:
         - name: DRIFT_RECONCILER_WORKERS
           value: {{ quote $.Values.agent.reconciler.workers.drift }}
         {{- end }}
+        {{- if $.Values.leaderElection.leaseDuration }}
+        - name: CATTLE_ELECTION_LEASE_DURATION
+          value: {{$.Values.leaderElection.leaseDuration}}
+        {{- end }}
+        {{- if $.Values.leaderElection.retryPeriod }}
+        - name: CATTLE_ELECTION_RETRY_PERIOD
+          value: {{$.Values.leaderElection.retryPeriod}}
+        {{- end }}
+        {{- if $.Values.leaderElection.renewDeadline }}
+        - name: CATTLE_ELECTION_RENEW_DEADLINE
+          value: {{$.Values.leaderElection.renewDeadline}}
+        {{- end }}
         image: '{{ template "system_default_registry" . }}{{.Values.image.repository}}:{{.Values.image.tag}}'
         name: fleet-agent
         command:
@@ -79,6 +102,18 @@ spec:
           valueFrom:
             fieldRef:
               fieldPath: metadata.namespace
+        {{- if $.Values.leaderElection.leaseDuration }}
+        - name: CATTLE_ELECTION_LEASE_DURATION
+          value: {{$.Values.leaderElection.leaseDuration}}
+        {{- end }}
+        {{- if $.Values.leaderElection.retryPeriod }}
+        - name: CATTLE_ELECTION_RETRY_PERIOD
+          value: {{$.Values.leaderElection.retryPeriod}}
+        {{- end }}
+        {{- if $.Values.leaderElection.renewDeadline }}
+        - name: CATTLE_ELECTION_RENEW_DEADLINE
+          value: {{$.Values.leaderElection.renewDeadline}}
+        {{- end }}
         image: '{{ template "system_default_registry" . }}{{.Values.image.repository}}:{{.Values.image.tag}}'
         name: fleet-agent-clusterstatus
         command:

charts/fleet-agent/values.yaml

@@ -81,3 +81,9 @@ agent:
     workers:
       bundledeployment: "50"
       drift: "50"
+
+## Leader election configuration
+leaderElection:
+  leaseDuration: 30s
+  retryPeriod: 10s
+  renewDeadline: 25s

dev/setup-fleet

@@ -50,7 +50,7 @@ helm -n cattle-fleet-system upgrade --install --create-namespace --wait --reset-
 # wait for controller and agent rollout
 kubectl -n cattle-fleet-system rollout status deployment/fleet-controller
 { grep -E -q -m 1 "fleet-agent-local.*1/1"; kill $!; } < <(kubectl get bundles -n fleet-local -w)
-kubectl -n cattle-fleet-system rollout status statefulset/fleet-agent
+kubectl -n cattle-fleet-system rollout status deployment/fleet-agent
 
 # label local cluster
 kubectl patch clusters.fleet.cattle.io -n fleet-local local --type=json -p '[{"op": "add", "path": "/metadata/labels/management.cattle.io~1cluster-display-name", "value": "local" }]'

e2e/installation/verify_test.go

@@ -63,15 +63,15 @@ var _ = Describe("Fleet Installation", func() {
 	Context("Verify agents are updated to new image", func() {
 		It("has the expected fleet images", func() {
 			Eventually(func() string {
-				out, _ := k.Namespace(localAgentNamespace).Get("statefulsets", "-owide")
+				out, _ := k.Namespace(localAgentNamespace).Get("deployments", "-owide")
 				return out
 			}).Should(ContainSubstring("rancher/fleet-agent:" + version))
 		})
 
 		It("has the expected fleet-agent image in the downstream cluster", Label("multi-cluster"), func() {
 			kd := env.Kubectl.Context(env.Downstream)
 			Eventually(func() string {
-				out, _ := kd.Namespace(agentNamespace).Get("statefulsets", "-owide")
+				out, _ := kd.Namespace(agentNamespace).Get("deployments", "-owide")
 				return out
 			}).Should(ContainSubstring("rancher/fleet-agent:" + version))
 		})

e2e/multi-cluster/installation/agent_test.go

@@ -5,18 +5,24 @@ import (
 
 	. "github.com/onsi/ginkgo/v2"
 	. "github.com/onsi/gomega"
-	"github.com/onsi/gomega/matchers"
-	"github.com/rancher/fleet/e2e/testenv/kubectl"
 )
 
 var (
 	agentMode string
-	kd        kubectl.Command
 )
 
 var _ = Describe("Fleet installation with TLS agent modes", func() {
 	BeforeEach(func() {
 		kd = env.Kubectl.Context(env.Downstream)
+
+		_, err := kd.Delete(
+			"pod",
+			"-n",
+			"cattle-fleet-system",
+			"-l",
+			"app=fleet-agent",
+		)
+		Expect(err).NotTo(HaveOccurred())
 	})
 
 	JustBeforeEach(func() {
@@ -33,7 +39,6 @@ var _ = Describe("Fleet installation with TLS agent modes", func() {
 			),
 		)
 		Expect(err).ToNot(HaveOccurred(), string(out))
-
 	})
 
 	Context("with non-strict agent TLS mode", func() {
@@ -43,7 +48,7 @@ var _ = Describe("Fleet installation with TLS agent modes", func() {
 			})
 
 			It("reaches the server without cert issues", func() {
-				Eventually(func() bool {
+				Eventually(func(g Gomega) error {
 					logs, err := kd.Namespace("cattle-fleet-system").Logs(
 						"-l",
 						"app=fleet-agent",
@@ -52,19 +57,13 @@ var _ = Describe("Fleet installation with TLS agent modes", func() {
 						"--tail=-1",
 					)
 					if err != nil {
-						return false
+						return err
 					}
 
-					regexMatcher := matchers.MatchRegexpMatcher{
-						Regexp: "Failed to register agent.*could not find the requested resource",
-					}
-					reachesServerWithoutCertIssue, err := regexMatcher.Match(logs)
-					if err != nil {
-						return false
-					}
+					g.Expect(logs).To(MatchRegexp("Failed to register agent.*could not find the requested resource"))
 
-					return reachesServerWithoutCertIssue
-				}).Should(BeTrue())
+					return nil
+				}).ShouldNot(HaveOccurred())
 			})
 		})
 	})
@@ -76,7 +75,7 @@ var _ = Describe("Fleet installation with TLS agent modes", func() {
 			})
 
 			It("cannot reach the server because the cert is signed by an unknown authority", func() {
-				Eventually(func() bool {
+				Eventually(func(g Gomega) error {
 					logs, err := kd.Namespace("cattle-fleet-system").Logs(
 						"-l",
 						"app=fleet-agent",
@@ -85,19 +84,16 @@ var _ = Describe("Fleet installation with TLS agent modes", func() {
 						"--tail=-1",
 					)
 					if err != nil {
-						return false
+						return err
 					}
 
-					regexMatcher := matchers.MatchRegexpMatcher{
-						Regexp: "Failed to register agent.*signed by unknown authority",
-					}
-					reachesServerWithoutCertIssue, err := regexMatcher.Match(logs)
+					g.Expect(logs).To(MatchRegexp("Failed to register agent.*signed by unknown authority"))
 					if err != nil {
-						return false
+						return err
 					}
 
-					return reachesServerWithoutCertIssue
-				}).Should(BeTrue())
+					return nil
+				}).ShouldNot(HaveOccurred())
 			})
 		})
 	})

e2e/multi-cluster/installation/suite_test.go

@@ -20,9 +20,11 @@ func TestE2E(t *testing.T) {
 }
 
 var (
-	env    *testenv.Env
-	ku     kubectl.Command
-	config string
+	env      *testenv.Env
+	ku       kubectl.Command
+	kd       kubectl.Command
+	config   string
+	strategy string
 )
 
 var _ = BeforeSuite(func() {
@@ -31,6 +33,7 @@ var _ = BeforeSuite(func() {
 
 	env = testenv.New()
 	ku = env.Kubectl.Context(env.Upstream)
+	kd = env.Kubectl.Context(env.Downstream)
 
 	// Save initial state of `fleet-controller` config map
 	cfg, err := ku.Get(
@@ -42,6 +45,29 @@ var _ = BeforeSuite(func() {
 		"jsonpath={.data.config}")
 	Expect(err).ToNot(HaveOccurred(), cfg)
 
+	// Save initial state of `fleet-agent` deployment
+	strategy, err = kd.Get(
+		"deployment",
+		"fleet-agent",
+		"-n",
+		"cattle-fleet-system",
+		"-o",
+		"jsonpath={.spec.strategy}",
+	)
+	Expect(err).ToNot(HaveOccurred(), cfg)
+
+	// Patch `fleet-agent` deployment to use Recreate strategy
+	out, err := kd.Patch(
+		"deployment",
+		"fleet-agent",
+		"-n",
+		"cattle-fleet-system",
+		"--type=merge",
+		"-p",
+		`{"spec":{"strategy":{"type":"Recreate", "rollingUpdate":null}}}`,
+	)
+	Expect(err).ToNot(HaveOccurred(), string(out))
+
 	cfg = strings.ReplaceAll(cfg, `"`, `\"`)
 	config = strings.ReplaceAll(cfg, "\n", "")
 })
@@ -58,4 +84,16 @@ var _ = AfterSuite(func() {
 		fmt.Sprintf(`{"data":{"config":"%s"}}`, config),
 	)
 	Expect(err).ToNot(HaveOccurred(), string(out))
+
+	// Restore initial state of deployment
+	out, err = kd.Patch(
+		"deployment",
+		"fleet-agent",
+		"-n",
+		"cattle-fleet-system",
+		"--type=merge",
+		"-p",
+		fmt.Sprintf(`{"spec":{"strategy":%s}}`, strategy),
+	)
+	Expect(err).ToNot(HaveOccurred(), string(out))
 })