Review changes

Signed-off-by: Sebastian Woehrl <[email protected]>

docs/userguide/main.md

@@ -685,7 +685,7 @@ The operator contains several features that automate management tasks that might
 ### Cluster recovery
 
 This operator automatically handles common failure scenarios and restarts crashed pods, normally this is done in a one-by-one fashion to maintain quorum and cluster stability.
-In case the operator detects several crashed or missing pods (for a nodepool) at the same time it will switch into a special recovery mode and start all pods at once and allow the cluster to form a new quorum. This parallel recovery mode is currently experimental and only works with PVC-backed storage. If you encounter problems with it, you can disable it by redeploying the operator and adding `manager.parallelRecoveryEnabled: false` to your `values.yaml`. Please also report any problems by opening an issue in the operator github project.
+In case the operator detects several crashed or missing pods (for a nodepool) at the same time it will switch into a special recovery mode and start all pods at once and allow the cluster to form a new quorum. This parallel recovery mode is currently experimental and only works with PVC-backed storage as it uses the number of existing PVCs to determine the number of missing pods. The recovery is done by temporarily changing the statefulset underlying each nodepool and setting the `podManagementPolicy` to `Parallel`. If you encounter problems with it, you can disable it by redeploying the operator and adding `manager.parallelRecoveryEnabled: false` to your `values.yaml`. Please also report any problems by opening an issue in the operator github project.
 
 The recovery mode also kicks in if you deleted your cluster but kept the PVCs around and are then reinstalling the cluster.
 

opensearch-operator/pkg/helpers/helpers.go

@@ -17,6 +17,11 @@ import (
 	"sigs.k8s.io/controller-runtime/pkg/client"
 )
 
+const (
+	stsUpdateWaitTime = 30
+	updateStepTime    = 3
+)
+
 func ContainsString(slice []string, s string) bool {
 	for _, item := range slice {
 		if item == s {
@@ -191,43 +196,43 @@ func WaitForSTSDelete(ctx context.Context, k8sClient client.Client, obj *appsv1.
 	if err := k8sClient.Delete(ctx, obj, &opts); err != nil {
 		return err
 	}
-	for i := 1; i <= 10; i++ {
+	for i := 1; i <= stsUpdateWaitTime/updateStepTime; i++ {
 		existing := appsv1.StatefulSet{}
 		err := k8sClient.Get(ctx, client.ObjectKeyFromObject(obj), &existing)
 		if err != nil {
 			return nil
 		}
-		time.Sleep(time.Second * 2)
+		time.Sleep(time.Second * updateStepTime)
 	}
 	return fmt.Errorf("failed to delete STS")
 }
 
 // Wait for max 30s until a STS has at least the given number of replicas
 func WaitForSTSReplicas(ctx context.Context, k8sClient client.Client, obj *appsv1.StatefulSet, replicas int32) error {
-	for i := 1; i <= 10; i++ {
+	for i := 1; i <= stsUpdateWaitTime/updateStepTime; i++ {
 		existing := appsv1.StatefulSet{}
 		err := k8sClient.Get(ctx, client.ObjectKeyFromObject(obj), &existing)
 		if err == nil {
 			if existing.Status.Replicas >= replicas {
 				return nil
 			}
 		}
-		time.Sleep(time.Second * 3)
+		time.Sleep(time.Second * updateStepTime)
 	}
 	return fmt.Errorf("failed to wait for replicas")
 }
 
 // Wait for max 30s until a STS has a normal status (CurrentRevision != "")
 func WaitForSTSStatus(ctx context.Context, k8sClient client.Client, obj *appsv1.StatefulSet) (*appsv1.StatefulSet, error) {
-	for i := 1; i <= 10; i++ {
+	for i := 1; i <= stsUpdateWaitTime/updateStepTime; i++ {
 		existing := appsv1.StatefulSet{}
 		err := k8sClient.Get(ctx, client.ObjectKeyFromObject(obj), &existing)
 		if err == nil {
 			if existing.Status.CurrentRevision != "" {
 				return &existing, nil
 			}
 		}
-		time.Sleep(time.Second * 3)
+		time.Sleep(time.Second * updateStepTime)
 	}
 	return nil, fmt.Errorf("failed to wait for STS")
 }