Implement leader election

Signed-off-by: Markus Blaschke <[email protected]>

Author: mblaschke

Dockerfile

@@ -19,7 +19,8 @@ RUN ./kube-pool-manager --help
 # FINAL IMAGE
 #############################################
 FROM gcr.io/distroless/base
-ENV LOG_JSON=1
+ENV LOG_JSON=1 \
+    LEASE_ENABLE=1
 COPY --from=build /go/src/github.com/webdevops/kube-pool-manager/kube-pool-manager /
 USER 1000
 ENTRYPOINT ["/kube-pool-manager"]

config/opts.go

@@ -15,11 +15,24 @@ type (
 			LogJson bool `           long:"log.json"     env:"LOG_JSON" description:"Switch log output to json format"`
 		}
 
+		// instance
+		Instance struct {
+			Nodename  *string `long:"instance.nodename"    env:"INSTANCE_NODENAME"   description:"Name of node where autopilot is running"`
+			Namespace *string `long:"instance.namespace"   env:"INSTANCE_NAMESPACE"   description:"Name of namespace where autopilot is running"`
+			Pod       *string `long:"instance.pod"         env:"INSTANCE_POD"         description:"Name of pod where autopilot is running"`
+		}
+
 		K8s struct {
 			NodeLabelSelector string        `long:"kube.node.labelselector"     env:"KUBE_NODE_LABELSELECTOR"     description:"Node Label selector which nodes should be checked"        default:""`
 			WatchTimeout      time.Duration `long:"kube.watch.timeout"          env:"KUBE_WATCH_TIMEOUT"          description:"Timeout & full resync for node watch (time.Duration)"     default:"24h"`
 		}
 
+		// lease
+		Lease struct {
+			Enabled bool   `long:"lease.enable"  env:"LEASE_ENABLE"  description:"Enable lease (leader election; enabled by default in docker images)"`
+			Name    string `long:"lease.name"    env:"LEASE_NAME"    description:"Name of lease lock"     default:"kube-pool-manager-leader"`
+		}
+
 		// general options
 		DryRun     bool   `long:"dry-run"  env:"DRY_RUN"       description:"Dry run (do not apply to nodes)"`
 		Config     string `long:"config"   env:"CONFIG"        description:"Config path"        required:"true"`

deployment/deployment.yaml

@@ -7,7 +7,7 @@ metadata:
   labels:
     app: kube-pool-manager
 spec:
-  replicas: 1
+  replicas: 2
   selector:
     matchLabels:
       app: kube-pool-manager
@@ -28,6 +28,19 @@ spec:
           env:
             - name: CONFIG
               value: "/config/pools.yaml"
+            # Instance
+            - name: INSTANCE_NODENAME
+              valueFrom:
+                fieldRef:
+                  fieldPath: spec.nodeName
+            - name: INSTANCE_POD
+              valueFrom:
+                fieldRef:
+                  fieldPath: metadata.name
+            - name: INSTANCE_NAMESPACE
+              valueFrom:
+                fieldRef:
+                  fieldPath: metadata.namespace
           securityContext:
             readOnlyRootFilesystem: true
             runAsNonRoot: true

deployment/rbac.yaml

@@ -9,6 +9,23 @@ rules:
     verbs:     ["get", "list", "patch", "watch"]
 ---
 apiVersion: rbac.authorization.k8s.io/v1
+kind: Role
+metadata:
+  name: kube-pool-manager
+  namespace: kube-system
+rules:
+  - apiGroups: [""]
+    resources: ["configmaps"]
+    verbs: ["create"]
+  - apiGroups: [""]
+    resources: ["configmaps"]
+    resourceNames: ["kube-pool-manager-leader"]
+    verbs: ["get", "watch", "update", "patch"]
+  - apiGroups: [""]
+    resources: ["pods"]
+    verbs:     ["get"]
+---
+apiVersion: rbac.authorization.k8s.io/v1
 kind: ClusterRoleBinding
 metadata:
   name: kube-pool-manager

go.mod

@@ -4,10 +4,12 @@ go 1.15
 
 require (
 	github.com/jessevdk/go-flags v1.4.0
+	github.com/operator-framework/operator-sdk v0.8.2
 	github.com/prometheus/client_golang v1.7.1
 	github.com/sirupsen/logrus v1.6.0
-	gopkg.in/yaml.v2 v2.2.8
-	k8s.io/api v0.18.0
-	k8s.io/apimachinery v0.18.0
-	k8s.io/client-go v0.18.0
+	gopkg.in/yaml.v2 v2.3.0
+	k8s.io/api v0.18.6
+	k8s.io/apimachinery v0.18.6
+	k8s.io/client-go v0.18.6
+	sigs.k8s.io/controller-runtime v0.6.2 // indirect
 )