Merge pull request #18490 from akka/wip-16948-singleton-retries-patriknw

=clt #16948 Use min retries for singleton leaving scenario

akka-cluster-tools/src/main/resources/reference.conf

@@ -141,6 +141,11 @@ akka.cluster.singleton {
   # the previous oldest confirms that the hand over has started or the previous 
   # oldest member is removed from the cluster (+ akka.cluster.down-removal-margin).
   hand-over-retry-interval = 1s
+  
+  # The number of retries are derived from hand-over-retry-interval and
+  # akka.cluster.down-removal-margin (or ClusterSingletonManagerSettings.removalMargin),
+  # but it will never be less than this property.
+  min-number-of-hand-over-retries = 10
 }
 # //#singleton-config
 

akka-cluster-tools/src/main/scala/akka/cluster/singleton/ClusterSingletonManager.scala

@@ -387,7 +387,13 @@ class ClusterSingletonManager(
 
   val (maxHandOverRetries, maxTakeOverRetries) = {
     val n = (removalMargin.toMillis / handOverRetryInterval.toMillis).toInt
-    (n + 3, math.max(1, n - 3))
+    val minRetries = context.system.settings.config.getInt(
+      "akka.cluster.singleton.min-number-of-hand-over-retries")
+    require(minRetries >= 1, "min-number-of-hand-over-retries must be >= 1")
+    val handOverRetries = math.max(minRetries, n + 3)
+    val takeOverRetries = math.max(1, handOverRetries - 3)
+
+    (handOverRetries, takeOverRetries)
   }
 
   // started when when self member is Up

akka-cluster-tools/src/multi-jvm/scala/akka/cluster/singleton/ClusterSingletonManagerSpec.scala

@@ -115,15 +115,21 @@ object ClusterSingletonManagerSpec extends MultiNodeConfig {
   /**
    * The Singleton actor
    */
-  class Consumer(queue: ActorRef, delegateTo: ActorRef) extends Actor {
+  class Consumer(queue: ActorRef, delegateTo: ActorRef) extends Actor with ActorLogging {
 
     import Consumer._
     import PointToPointChannel._
 
     var current = 0
+    var stoppedBeforeUnregistration = true
 
     override def preStart(): Unit = queue ! RegisterConsumer
 
+    override def postStop(): Unit = {
+      if (stoppedBeforeUnregistration)
+        log.warning("Stopped before unregistration")
+    }
+
     def receive = {
       case n: Int if n <= current ⇒
         context.stop(self)
@@ -138,6 +144,7 @@ object ClusterSingletonManagerSpec extends MultiNodeConfig {
       case End ⇒
         queue ! UnregisterConsumer
       case UnregistrationOk ⇒
+        stoppedBeforeUnregistration = false
         context stop self
       case Ping ⇒
         sender() ! Pong