humio · SaaldjorMike · Mar 27, 2020 · Mar 26, 2020 · Mar 26, 2020 · Mar 26, 2020
diff --git a/deploy/crds/core.humio.com_humioclusters_crd.yaml b/deploy/crds/core.humio.com_humioclusters_crd.yaml
@@ -3,6 +3,19 @@ kind: CustomResourceDefinition
 metadata:
   name: humioclusters.core.humio.com
 spec:
+  additionalPrinterColumns:
+  - JSONPath: .status.clusterState
+    description: The state of the cluster
+    name: State
+    type: string
+  - JSONPath: .status.clusterNodeCount
+    description: The number of nodes in the cluster
+    name: Nodes
+    type: string
+  - JSONPath: .status.clusterVersion
+    description: The version of humior
+    name: Version
+    type: string
   group: core.humio.com
   names:
     kind: HumioCluster
@@ -151,10 +164,15 @@ spec:
             allDataAvailable:
               description: Current state set by operator.
               type: string
+            clusterNodeCount:
+              description: ClusterNodeCount is the number of nodes of humio running
+              type: integer
             clusterState:
               description: 'ClusterState will be empty before the cluster is bootstrapped.
-                From there it can be "Bootstrapping" or "Operational" TODO: other
-                states?'
+                From there it can be "Bootstrapping" or "Running" TODO: other states?'
+              type: string
+            clusterVersion:
+              description: ClusterVersion is the version of humio running
               type: string
             stateLastUpdated:
               format: int64

diff --git a/pkg/apis/core/v1alpha1/humiocluster_types.go b/pkg/apis/core/v1alpha1/humiocluster_types.go
@@ -5,6 +5,13 @@ import (
 	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
 )
 
+const (
+	// HumioClusterStateBoostrapping is the Bootstrapping state of the cluster
+	HumioClusterStateBoostrapping = "Bootstrapping"
+	// HumioClusterStateRunning is the Running state of the cluster
+	HumioClusterStateRunning = "Running"
+)
+
 // HumioClusterSpec defines the desired state of HumioCluster
 type HumioClusterSpec struct {
 	// Desired container image including the image tag
@@ -26,16 +33,23 @@ type HumioClusterStatus struct {
 	StateLastUpdatedUnix int64 `json:"stateLastUpdated,omitempty"`
 	// Current state set by operator.
 	AllDataAvailable string `json:"allDataAvailable,omitempty"`
-	// ClusterState will be empty before the cluster is bootstrapped. From there it can be "Bootstrapping" or "Operational"
+	// ClusterState will be empty before the cluster is bootstrapped. From there it can be "Bootstrapping" or "Running"
 	// TODO: other states?
 	ClusterState string `json:"clusterState,omitempty"`
+	// ClusterVersion is the version of humio running
+	ClusterVersion string `json:"clusterVersion,omitempty"`
+	// ClusterNodeCount is the number of nodes of humio running
+	ClusterNodeCount int `json:"clusterNodeCount,omitempty"`
 }
 
 // +k8s:deepcopy-gen:interfaces=k8s.io/apimachinery/pkg/runtime.Object
 
 // HumioCluster is the Schema for the humioclusters API
 // +kubebuilder:subresource:status
 // +kubebuilder:resource:path=humioclusters,scope=Namespaced
+// +kubebuilder:printcolumn:name="State",type="string",JSONPath=".status.clusterState",description="The state of the cluster"
+// +kubebuilder:printcolumn:name="Nodes",type="string",JSONPath=".status.clusterNodeCount",description="The number of nodes in the cluster"
+// +kubebuilder:printcolumn:name="Version",type="string",JSONPath=".status.clusterVersion",description="The version of humior"
 type HumioCluster struct {
 	metav1.TypeMeta   `json:",inline"`
 	metav1.ObjectMeta `json:"metadata,omitempty"`

diff --git a/pkg/controller/humiocluster/humiocluster_controller.go b/pkg/controller/humiocluster/humiocluster_controller.go
@@ -131,7 +131,7 @@ func (r *ReconcileHumioCluster) Reconcile(request reconcile.Request) (reconcile.
 	// Assume we are bootstrapping if no cluster state is set.
 	// TODO: this is a workaround for the issue where humio pods cannot start up at the same time during the first boot
 	if humioCluster.Status.ClusterState == "" {
-		r.setClusterStatus(context.TODO(), "Boostrapping", humioCluster)
+		r.setClusterState(context.TODO(), corev1alpha1.HumioClusterStateBoostrapping, humioCluster)
 	}
 
 	// Ensure developer password is a k8s secret
@@ -149,13 +149,34 @@ func (r *ReconcileHumioCluster) Reconcile(request reconcile.Request) (reconcile.
 	}
 
 	// Ensure pods exist. Will requeue if not all pods are created and ready
+	if humioCluster.Status.ClusterState == corev1alpha1.HumioClusterStateBoostrapping {
+		result, err = r.ensurePodsBootstrapped(context.TODO(), humioCluster)
+		if result != emptyResult || err != nil {
+			return result, err
+		}
+	}
+
+	r.setClusterState(context.TODO(), corev1alpha1.HumioClusterStateRunning, humioCluster)
+
+	defer func(context context.Context, humioCluster *corev1alpha1.HumioCluster) {
+		pods, _ := ListPods(r.client, humioCluster)
+		r.setClusterNodeCount(context, len(pods), humioCluster)
+	}(context.TODO(), humioCluster)
+
+	// TODO: get cluster version from humio api
+	defer func(context context.Context, humioClient humio.Client, humioCluster *corev1alpha1.HumioCluster) {
+		status, err := humioClient.Status()
+		if err != nil {
+			r.logger.Info("unable to get status: %s", err)
+		}
+		r.setClusterVersion(context, status.Version, humioCluster)
+	}(context.TODO(), r.humioClient, humioCluster)
+
 	result, err = r.ensurePodsExist(context.TODO(), humioCluster)
 	if result != emptyResult || err != nil {
 		return result, err
 	}
 
-	r.setClusterStatus(context.TODO(), "Running", humioCluster)
-
 	// Ensure service exists
 	err = r.ensureServiceExists(context.TODO(), humioCluster)
 	if err != nil {
@@ -173,6 +194,7 @@ func (r *ReconcileHumioCluster) Reconcile(request reconcile.Request) (reconcile.
 		return reconcile.Result{}, err
 	}
 
+	// TODO: wait until all pods are ready before continuing
 	clusterController := humio.NewClusterController(r.humioClient)
 	err = r.ensurePartitionsAreBalanced(*clusterController, humioCluster)
 	if err != nil {
@@ -183,11 +205,21 @@ func (r *ReconcileHumioCluster) Reconcile(request reconcile.Request) (reconcile.
 	return reconcile.Result{Requeue: true, RequeueAfter: time.Second * 30}, nil
 }
 
-// setClusterStatus is used to change the cluster status
+// setClusterState is used to change the cluster state
 // TODO: we use this to determine if we should have a delay between startup of humio pods during bootstrap vs starting up pods during an image update
-func (r *ReconcileHumioCluster) setClusterStatus(context context.Context, clusterState string, humioCluster *corev1alpha1.HumioCluster) error {
+func (r *ReconcileHumioCluster) setClusterState(context context.Context, clusterState string, humioCluster *corev1alpha1.HumioCluster) error {
 	humioCluster.Status.ClusterState = clusterState
-	return r.client.Update(context, humioCluster)
+	return r.client.Status().Update(context, humioCluster)
+}
+
+func (r *ReconcileHumioCluster) setClusterVersion(context context.Context, clusterVersion string, humioCluster *corev1alpha1.HumioCluster) error {
+	humioCluster.Status.ClusterVersion = clusterVersion
+	return r.client.Status().Update(context, humioCluster)
+}
+
+func (r *ReconcileHumioCluster) setClusterNodeCount(context context.Context, clusterNodeCount int, humioCluster *corev1alpha1.HumioCluster) error {
+	humioCluster.Status.ClusterNodeCount = clusterNodeCount
+	return r.client.Status().Update(context, humioCluster)
 }
 
 func (r *ReconcileHumioCluster) ensurePodLabels(context context.Context, hc *corev1alpha1.HumioCluster) error {
@@ -315,7 +347,7 @@ func (r *ReconcileHumioCluster) ensureMismatchedPodVersionsAreDeleted(conetext c
 
 // TODO: change to create 1 pod at a time, return Requeue=true and RequeueAfter.
 // check that other pods, if they exist, are in a ready state
-func (r *ReconcileHumioCluster) ensurePodsExist(conetext context.Context, humioCluster *corev1alpha1.HumioCluster) (reconcile.Result, error) {
+func (r *ReconcileHumioCluster) ensurePodsBootstrapped(conetext context.Context, humioCluster *corev1alpha1.HumioCluster) (reconcile.Result, error) {
 	// Ensure we have pods for the defined NodeCount.
 	// If scaling down, we will handle the extra/obsolete pods later.
 	foundPodList, err := ListPods(r.client, humioCluster)
@@ -341,7 +373,7 @@ func (r *ReconcileHumioCluster) ensurePodsExist(conetext context.Context, humioC
 		return reconcile.Result{}, nil
 	}
 
-	if podsNotReadyCount > 0 && humioCluster.Status.ClusterState == "Bootstrapping" {
+	if podsNotReadyCount > 0 {
 		r.logger.Info(fmt.Sprintf("there are %d humio pods that are not ready. all humio pods must report ready before reconciliation can continue", podsNotReadyCount))
 		return reconcile.Result{Requeue: true, RequeueAfter: time.Second * 5}, nil
 	}
@@ -367,6 +399,35 @@ func (r *ReconcileHumioCluster) ensurePodsExist(conetext context.Context, humioC
 	return reconcile.Result{}, nil
 }
 
+func (r *ReconcileHumioCluster) ensurePodsExist(conetext context.Context, humioCluster *corev1alpha1.HumioCluster) (reconcile.Result, error) {
+	// Ensure we have pods for the defined NodeCount.
+	// If scaling down, we will handle the extra/obsolete pods later.
+	foundPodList, err := ListPods(r.client, humioCluster)
+	if err != nil {
+		return reconcile.Result{}, fmt.Errorf("failed to list pods: %s", err)
+	}
+
+	if len(foundPodList) < humioCluster.Spec.NodeCount {
+		pod, err := r.constructPod(humioCluster)
+		if err != nil {
+			return reconcile.Result{}, fmt.Errorf("unable to construct pod for HumioCluster: %v", err)
+		}
+
+		err = r.client.Create(context.TODO(), pod)
+		if err != nil {
+			log.Info(fmt.Sprintf("unable to create pod: %v", err))
+			return reconcile.Result{Requeue: true, RequeueAfter: time.Second * 5}, fmt.Errorf("unable to create Pod for HumioCluster: %v", err)
+		}
+		log.Info(fmt.Sprintf("successfully created pod %s for HumioCluster %s", pod.Name, humioCluster.Name))
+		metricPodsCreated.Inc()
+		// We have created a pod. Requeue immediately even if the pod is not ready. We will check the readiness status on the next reconciliation.
+		return reconcile.Result{Requeue: true}, nil
+	}
+
+	// TODO: what should happen if we have more pods than are expected?
+	return reconcile.Result{}, nil
+}
+
 // TODO: extend this (or create separate method) to take this password and perform a login, get the jwt token and then call the api to get the persistent api token and also store that as a secret
 // this functionality should perhaps go into humio.cluster_auth.go
 func (r *ReconcileHumioCluster) ensureDeveloperUserPasswordExists(conetext context.Context, humioCluster *corev1alpha1.HumioCluster) error {