Akka.Cluster: enable keep-majority default SBR

docs/articles/clustering/split-brain-resolver.md

@@ -38,6 +38,19 @@ Keep in mind that split brain resolver will NOT work when `akka.cluster.auto-dow
 
 Beginning in Akka.NET v1.4.16, the Akka.NET project has ported the original split brain resolver implementations from Lightbend as they are now open source. The following section of documentation describes how Akka.NET's hand-rolled split brain resolvers are implemented.
 
+> [!IMPORTANT]
+> As of Akka.NET v1.5.2, the `keep-majority` split brain resolution strategy is now enabled by default. This should be acceptable for the majority of Akka.Cluster users, but please read on.
+
+### Disabling the Default Downing Provider
+
+To disable the default Akka.Cluster downing provider, simply configure the following in your HOCON:
+
+```hocon
+akka.cluster.downing-provider-class = ""
+```
+
+This will disable the split brain resolver / downing provider functionality altogether in Akka.NET. This was the default behavior for Akka.Cluster as of Akka.NET v1.5.1 and earlier.
+
 ### Picking a Strategy
 
 In order to enable an Akka.NET split brain resolver in your cluster (they are not enabled by default), you will want to update your `akka.cluster` HOCON configuration to the following:
@@ -59,7 +72,7 @@ This will cause the [`Akka.Cluster.SBR.SplitBrainResolverProvider`](xref:Akka.Cl
 The following strategies are supported:
 
 * `static-quorum`
-* `keep-majority`
+* `keep-majority` **(default)**
 * `keep-oldest`
 * `down-all`
 * `lease-majority`
@@ -144,6 +157,9 @@ akka.cluster.split-brain-resolver {
 
 #### Keep Majority
 
+> [!NOTE]
+> `keep-majority` is the default SBR strategy for Akka.Cluster as of Akka.NET v1.5.2+.
+
 The `keep-majority` strategy will down this part of the cluster, which sees a lesser part of the whole cluster. This choice is made based on the latest known state of the cluster. When cluster will split into two equal parts, the one which contains the lowest address, will survive.
 
 When to use it? When your cluster can grow or shrink very dynamically.

docs/community/whats-new/akkadotnet-v1.5-upgrade-advisories.md

@@ -11,6 +11,48 @@ This document contains specific upgrade suggestions, warnings, and notices that
 <iframe width="560" height="315" src="https://www.youtube.com/embed/-UPestlIw4k" title="YouTube video player" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share" allowfullscreen></iframe>
 <!-- markdownlint-enable MD033 -->
 
+## Upgrading to Akka.NET v1.5.2
+
+Akka.NET v1.5.2 introduces two important behavioral changes:
+
+* [Akka.Persistence: need to remove hard-coded Newtonsoft.Json `object` serializer](https://github.com/akkadotnet/akka.net/issues/6389)
+* [Akka.Cluster: enable `keep-majority` as default Split Brain Resolver](https://github.com/akkadotnet/akka.net/pull/6628)
+
+We meant to include both of these changes in Akka.NET v1.5.0 but simply ran out of time before making them into that release.
+
+### Akka.Persistence Changes
+
+The impact of [Akka.Persistence: need to remove hard-coded Newtonsoft.Json `object` serializer](https://github.com/akkadotnet/akka.net/issues/6389) is pretty minor: all versions of Akka.NET prior to 1.5.2 used Newtonsoft.Json as the `object` serializer for Akka.Persistence regardless of whether or not you [used a custom `object` serializer, such as Hyperion](xref:serialization#complex-object-serialization-using-hyperion).
+
+Going forward your user-defined `object` serialization binding will now be respected by Akka.Persistence. Any old data previously saved using Newtonsoft.Json will continue to be recovered automatically by Newtonsoft.Json - it's only the serialization of new objects inserted after upgrading to v1.5.2 that will be affected.
+
+If you _never changed your `object`_ serializer (most users don't) then this change doesn't affect you.
+
+### Akka.Cluster Split Brain Resolver Changes
+
+As of Akka.NET v1.5.2 we've now enabled the `keep-majority` [Split Brain Resolver](xref:split-brain-resolver) by default.
+
+If you were already running with a custom SBR enabled, this change won't affect you.
+
+If you weren't running with an SBR enabled, you should read the [Akka.Cluster Split Brain Resolver documentation](xref:split-brain-resolver).
+
+Also worth noting: we've disabled the `akka.cluster.auto-down-unreachable-after` setting as it's always been a poor and shoddy way to manage network partitions inside Akka.Cluster. If you have that setting enabled it will be ignored and you'll see the following warning appear instead:
+
+```shell
+The `auto-down-unreachable-after` feature has been deprecated as of Akka.NET v1.5.2 and will be removed in a future version of Akka.NET.
+The `keep-majority` split brain resolver will be used instead. See https://getakka.net/articles/cluster/split-brain-resolver.html for more details.
+```
+
+#### Disabling the Default Downing Provider
+
+To disable the default Akka.Cluster downing provider, simply configure the following in your HOCON:
+
+```hocon
+akka.cluster.downing-provider-class = ""
+```
+
+This will disable the split brain resolver / downing provider functionality altogether in Akka.NET. This was the default behavior for Akka.Cluster as of Akka.NET v1.5.1 and earlier.
+
 ## Upgrading From Akka.NET v1.4 to v1.5
 
 In case you need help upgrading:

src/contrib/cluster/Akka.Cluster.Sharding.Tests.MultiNode/ClusterShardingSpec.cs

@@ -58,7 +58,9 @@ public ClusterShardingSpecConfig(
             CommonConfig = ConfigurationFactory.ParseString($@"
                 akka.cluster.sharding.verbose-debug-logging = on
                 #akka.loggers = [""akka.testkit.SilenceAllTestEventListener""]
-
+                akka.cluster.downing-provider-class = ""Akka.Cluster.SBR.SplitBrainResolverProvider, Akka.Cluster""
+                akka.cluster.split-brain-resolver.stable-after = 1s
+                akka.cluster.down-removal-margin = 1s
                 akka.cluster.roles = [""backend""]
                 akka.cluster.distributed-data.gossip-interval = 1s
                 akka.persistence.journal.sqlite-shared.timeout = 10s #the original default, base test uses 5s

src/contrib/cluster/Akka.Cluster.Sharding.Tests.MultiNode/MultiNodeClusterShardingConfig.cs

@@ -58,7 +58,9 @@ protected MultiNodeClusterShardingConfig(
             Common =
                 ConfigurationFactory.ParseString($@"
                     akka.actor.provider = ""cluster""
-                    akka.cluster.auto-down-unreachable-after = 0s
+                    akka.cluster.downing-provider-class = ""Akka.Cluster.SBR.SplitBrainResolverProvider, Akka.Cluster""
+                    akka.cluster.split-brain-resolver.stable-after = 1s
+                    akka.cluster.down-removal-margin = 1s
                     akka.cluster.sharding.state-store-mode = ""{mode}""
                     akka.cluster.sharding.remember-entities = {rememberEntities.ToString().ToLowerInvariant()}
                     akka.cluster.sharding.remember-entities-store = ""{rememberEntitiesStore}""

src/contrib/cluster/Akka.Cluster.Tools.Tests/Singleton/ClusterSingletonConfigSpec.cs

@@ -37,7 +37,7 @@ public void ClusterSingletonManagerSettings_must_have_default_config()
             clusterSingletonManagerSettings.SingletonName.ShouldBe("singleton");
             clusterSingletonManagerSettings.Role.ShouldBe(null);
             clusterSingletonManagerSettings.HandOverRetryInterval.TotalSeconds.ShouldBe(1);
-            clusterSingletonManagerSettings.RemovalMargin.TotalSeconds.ShouldBe(0);
+            clusterSingletonManagerSettings.RemovalMargin.TotalSeconds.ShouldBe(20); // now 20 due to default SBR settings
 
             var config = Sys.Settings.Config.GetConfig("akka.cluster.singleton");
             Assert.False(config.IsNullOrEmpty());

src/core/Akka.API.Tests/verify/CoreAPISpec.ApproveCluster.DotNet.verified.txt

@@ -18,6 +18,7 @@
 [assembly: System.Runtime.Versioning.TargetFrameworkAttribute(".NETCoreApp,Version=v6.0", FrameworkDisplayName=".NET 6.0")]
 namespace Akka.Cluster
 {
+    [System.ObsoleteAttribute("No longer used as of Akka.NET v1.5.2")]
     public sealed class AutoDowning : Akka.Cluster.IDowningProvider
     {
         public AutoDowning(Akka.Actor.ActorSystem system, Akka.Cluster.Cluster cluster) { }
@@ -192,6 +193,8 @@ namespace Akka.Cluster
         public ClusterSettings(Akka.Configuration.Config config, string systemName) { }
         public bool AllowWeaklyUpMembers { get; }
         public Akka.Util.AppVersion AppVersion { get; }
+        [System.ObsoleteAttribute("No longer used as of Akka.NET v1.5.2 - clustering defaults to using KeepMajority " +
+            "SBR instead")]
         public System.Nullable<System.TimeSpan> AutoDownUnreachableAfter { get; }
         public System.Type DowningProviderType { get; }
         public Akka.Configuration.Config FailureDetectorConfig { get; }

src/core/Akka.API.Tests/verify/CoreAPISpec.ApproveCluster.Net.verified.txt

@@ -18,6 +18,7 @@
 [assembly: System.Runtime.Versioning.TargetFrameworkAttribute(".NETStandard,Version=v2.0", FrameworkDisplayName=".NET Standard 2.0")]
 namespace Akka.Cluster
 {
+    [System.ObsoleteAttribute("No longer used as of Akka.NET v1.5.2")]
     public sealed class AutoDowning : Akka.Cluster.IDowningProvider
     {
         public AutoDowning(Akka.Actor.ActorSystem system, Akka.Cluster.Cluster cluster) { }
@@ -192,6 +193,8 @@ namespace Akka.Cluster
         public ClusterSettings(Akka.Configuration.Config config, string systemName) { }
         public bool AllowWeaklyUpMembers { get; }
         public Akka.Util.AppVersion AppVersion { get; }
+        [System.ObsoleteAttribute("No longer used as of Akka.NET v1.5.2 - clustering defaults to using KeepMajority " +
+            "SBR instead")]
         public System.Nullable<System.TimeSpan> AutoDownUnreachableAfter { get; }
         public System.Type DowningProviderType { get; }
         public Akka.Configuration.Config FailureDetectorConfig { get; }

src/core/Akka.Cluster.TestKit/MultiNodeClusterSpec.cs

@@ -59,6 +59,7 @@ public static Config ClusterConfig()
                         retry-interval = 200ms
                         waiting-for-state-timeout = 200ms
                     }
+                    #downing-provider-class = """" # disable default SBR
                 }
                 akka.loglevel = INFO
                 akka.log-dead-letters = off

src/core/Akka.Cluster.Tests/ClusterConfigSpec.cs

@@ -8,6 +8,7 @@
 using System;
 using System.Collections.Immutable;
 using Akka.Actor;
+using Akka.Cluster.SBR;
 using Akka.Configuration;
 using Akka.Dispatch;
 using Akka.Remote;
@@ -44,7 +45,9 @@ public void Clustering_must_be_able_to_parse_generic_cluster_config_elements()
             settings.AllowWeaklyUpMembers.Should().BeTrue();
             settings.WeaklyUpAfter.Should().Be(7.Seconds());
             settings.PublishStatsInterval.Should().NotHaveValue();
+#pragma warning disable CS0618
             settings.AutoDownUnreachableAfter.Should().NotHaveValue();
+#pragma warning restore CS0618
             settings.DownRemovalMargin.Should().Be(TimeSpan.Zero);
             settings.MinNrOfMembers.Should().Be(1);
             settings.MinNrOfMembersOfRole.Should().Equal(ImmutableDictionary<string, int>.Empty);
@@ -71,6 +74,13 @@ public void Clustering_must_be_able_to_parse_generic_cluster_config_elements()
             settings.VerboseHeartbeatLogging.Should().BeFalse();
             settings.VerboseGossipReceivedLogging.Should().BeFalse();
             settings.RunCoordinatedShutdownWhenDown.Should().BeTrue();
+
+            // downing provider settings
+            settings.DowningProviderType.Should().Be<SplitBrainResolverProvider>();
+            var sbrSettings = new SplitBrainResolverSettings(Sys.Settings.Config);
+            sbrSettings.DowningStableAfter.Should().Be(20.Seconds());
+            sbrSettings.DownAllWhenUnstable.Should().Be(15.Seconds()); // 3/4 OF DowningStableAfter
+            sbrSettings.DowningStrategy.Should().Be("keep-majority");
         }
 
         /// <summary>
@@ -83,5 +93,17 @@ public void Clustering_should_parse_nondefault_AppVersion()
             var settings = new ClusterSettings(config.WithFallback(Sys.Settings.Config), Sys.Name);
             settings.AppVersion.Should().Be(AppVersion.Zero);
         }
+
+        /// <summary>
+        /// Validate that we can disable the default downing provider if needed
+        /// </summary>
+        [Fact]
+        public void Cluster_should_allow_disabling_of_default_DowningProvider()
+        {
+            // configure HOCON to disable the default akka.cluster downing provider
+            Config config = "akka.cluster.downing-provider-class = \"\"";
+            var settings = new ClusterSettings(config.WithFallback(Sys.Settings.Config), Sys.Name);
+            settings.DowningProviderType.Should().Be<NoDowning>();
+        }
     }
 }

src/core/Akka.Cluster.Tests/DowningProviderSpec.cs

@@ -34,7 +34,7 @@ public Props DowningActorProps
         }
     }
 
-    class DummyDowningProvider : IDowningProvider
+    internal class DummyDowningProvider : IDowningProvider
     {
         public readonly AtomicBoolean ActorPropsAccessed = new AtomicBoolean(false);
         public DummyDowningProvider(ActorSystem system, Cluster cluster)
@@ -69,21 +69,21 @@ public class DowningProviderSpec : AkkaSpec
         ");
 
         [Fact]
-        public void Downing_provider_should_default_to_NoDowning()
+        public void Downing_provider_should_default_to_KeepMajority()
         {
             using (var system = ActorSystem.Create("default", BaseConfig))
             {
-                Cluster.Get(system).DowningProvider.Should().BeOfType<NoDowning>();
+                Cluster.Get(system).DowningProvider.Should().BeOfType<Akka.Cluster.SBR.SplitBrainResolverProvider>();
             }
         }
 
         [Fact]
-        public void Downing_provider_should_use_AutoDowning_if_auto_down_unreachable_after_is_configured()
+        public void Downing_provider_should_ignore_AutoDowning_if_auto_down_unreachable_after_is_configured()
         {
             var config = ConfigurationFactory.ParseString(@"akka.cluster.auto-down-unreachable-after=18s");
             using (var system = ActorSystem.Create("auto-downing", config.WithFallback(BaseConfig)))
             {
-                Cluster.Get(system).DowningProvider.Should().BeOfType<AutoDowning>();
+                Cluster.Get(system).DowningProvider.Should().BeOfType<Akka.Cluster.SBR.SplitBrainResolverProvider>();
             }
         }
 
@@ -97,7 +97,7 @@ public void Downing_provider_should_use_specified_downing_provider()
                 var downingProvider = Cluster.Get(system).DowningProvider;
                 downingProvider.Should().BeOfType<DummyDowningProvider>();
                 AwaitCondition(() =>
-                    (downingProvider as DummyDowningProvider).ActorPropsAccessed.Value,
+                    ((DummyDowningProvider)downingProvider).ActorPropsAccessed.Value,
                     TimeSpan.FromSeconds(3));
             }
         }

src/core/Akka.Cluster/AutoDown.cs

@@ -270,6 +270,7 @@ private void Remove(UniqueAddress node)
     /// <summary>
     /// Used when no custom provider is configured and 'auto-down-unreachable-after' is enabled.
     /// </summary>
+    [Obsolete("No longer used as of Akka.NET v1.5.2")]
     public sealed class AutoDowning : IDowningProvider
     {
         private readonly ActorSystem _system;
@@ -296,7 +297,9 @@ public Props DowningActorProps
         {
             get
             {
+#pragma warning disable CS0618 // disable obsolete warning here because this entire class is obsolete
                 var autoDownUnreachableAfter = _cluster.Settings.AutoDownUnreachableAfter;
+#pragma warning restore CS0618
                 if (!autoDownUnreachableAfter.HasValue)
                     throw new ConfigurationException("AutoDowning downing provider selected but 'akka.cluster.auto-down-unreachable-after' not set");
 

src/core/Akka.Cluster/Cluster.cs

@@ -67,13 +67,11 @@ static Cluster()
             bool GetAssertInvariants()
             {
                 var isOn = Environment.GetEnvironmentVariable("AKKA_CLUSTER_ASSERT")?.ToLowerInvariant();
-                switch (isOn)
+                return isOn switch
                 {
-                    case "on":
-                        return true;
-                    default:
-                        return false;
-                }
+                    "on" => true,
+                    _ => false
+                };
             }
 
             IsAssertInvariantsEnabled = GetAssertInvariants();
@@ -114,12 +112,24 @@ public Cluster(ActorSystemImpl system)
             System = system;
             Settings = new ClusterSettings(system.Settings.Config, system.Name);
 
-            if (!(system.Provider is IClusterActorRefProvider provider))
+            if (system.Provider is not IClusterActorRefProvider provider)
                 throw new ConfigurationException(
                     $"ActorSystem {system} needs to have a 'IClusterActorRefProvider' enabled in the configuration, currently uses {system.Provider.GetType().FullName}");
             SelfUniqueAddress = new UniqueAddress(provider.Transport.DefaultAddress, AddressUidExtension.Uid(system));
 
             _log = Logging.GetLogger(system, "Cluster");
+
+            // log a warning if the user has set auto-down-unreachable-after to any value other than "off"
+            // obsolete setting, so suppress obsolete warning
+#pragma warning disable CS0618
+            if (Settings.AutoDownUnreachableAfter != null)
+#pragma warning restore CS0618
+            {
+                _log.Warning(
+                    "The `auto-down-unreachable-after` feature has been deprecated as of Akka.NET v1.5.2 and will be removed in a future version of Akka.NET. " +
+                    "The `keep-majority` split brain resolver will be used instead. See https://getakka.net/articles/cluster/split-brain-resolver.html for more details.");
+            }
+
 
             CurrentInfoLogger = new InfoLogger(_log, Settings, SelfAddress);
 

src/core/Akka.Cluster/ClusterDaemon.cs

@@ -225,7 +225,7 @@ public override bool Equals(object obj)
             {
                 if (ReferenceEquals(null, obj)) return false;
                 if (ReferenceEquals(this, obj)) return true;
-                return obj is Welcome && Equals((Welcome)obj);
+                return obj is Welcome welcome && Equals(welcome);
             }
 
             private bool Equals(Welcome other)
@@ -290,6 +290,16 @@ public override bool Equals(object obj)
             {
                 return obj is InitJoin;
             }
+
+            protected bool Equals(InitJoin other)
+            {
+                return true;
+            }
+
+            public override int GetHashCode()
+            {
+                return 1;
+            }
         }
 
         /// <inheritdoc cref="JoinSeenNode"/>