Cluster: Proactively reconfigure when we hit a MOVED response

docs/ReleaseNotes.md

@@ -8,7 +8,7 @@ Current package versions:
 
 ## Unreleased
 
-No pending changes for the next release yet.
+- Fix [#1520](https://github.com/StackExchange/StackExchange.Redis/issues/1520) & [#1660](https://github.com/StackExchange/StackExchange.Redis/issues/1660): When `MOVED` is encountered from a cluster, a reconfigure will happen proactively to react to cluster changes ASAP ([#2286 by NickCraver](https://github.com/StackExchange/StackExchange.Redis/pull/2286))
 
 
 ## 2.6.80

src/StackExchange.Redis/ConnectionMultiplexer.cs

@@ -49,6 +49,10 @@ public sealed partial class ConnectionMultiplexer : IInternalConnectionMultiplex
 
         ConfigurationOptions IInternalConnectionMultiplexer.RawConfig => RawConfig;
 
+        private int lastReconfigiureTicks = Environment.TickCount;
+        internal long LastReconfigureSecondsAgo =>
+            unchecked(Environment.TickCount - Thread.VolatileRead(ref lastReconfigiureTicks)) / 1000;
+
         private int _activeHeartbeatErrors, lastHeartbeatTicks;
         internal long LastHeartbeatSecondsAgo =>
             pulse is null
@@ -366,8 +370,19 @@ internal void CheckMessage(Message message)
             }
         }
 
-        internal bool TryResend(int hashSlot, Message message, EndPoint endpoint, bool isMoved) =>
-            ServerSelectionStrategy.TryResend(hashSlot, message, endpoint, isMoved);
+        internal bool TryResend(int hashSlot, Message message, EndPoint endpoint, bool isMoved)
+        {
+            // If we're being told to re-send something because the hash slot moved, that means our topology is out of date
+            // ...and we should re-evaluate what's what.
+            // Allow for a 5-second back-off so we don't hammer this in a loop though
+            if (isMoved && LastReconfigureSecondsAgo > 5)
+            {
+                // Async kickoff a reconfigure
+                ReconfigureIfNeeded(endpoint, false, "MOVED encountered");
+            }
+
+            return ServerSelectionStrategy.TryResend(hashSlot, message, endpoint, isMoved);
+        }
 
         /// <summary>
         /// Wait for a given asynchronous operation to complete (or timeout).
@@ -1214,6 +1229,7 @@ internal async Task<bool> ReconfigureAsync(bool first, bool reconfigureAll, LogP
                 }
                 Trace("Starting reconfiguration...");
                 Trace(blame != null, "Blaming: " + Format.ToString(blame));
+                Interlocked.Exchange(ref lastReconfigiureTicks, Environment.TickCount);
 
                 log?.WriteLine(RawConfig.ToString(includePassword: false));
                 log?.WriteLine();
@@ -1552,10 +1568,7 @@ public EndPoint[] GetEndPoints(bool configuredOnly = false) =>
                 foreach (EndPoint endpoint in clusterEndpoints)
                 {
                     serverEndpoint = GetServerEndPoint(endpoint);
-                    if (serverEndpoint != null)
-                    {
-                        serverEndpoint.UpdateNodeRelations(clusterConfig);
-                    }
+                    serverEndpoint?.UpdateNodeRelations(clusterConfig);
                 }
                 return clusterEndpoints;
             }

tests/StackExchange.Redis.Tests/CommandTimeoutTests.cs

@@ -47,7 +47,7 @@ public async Task DefaultHeartbeatLowTimeout()
         using var conn = ConnectionMultiplexer.Connect(options);
 
         var pauseServer = GetServer(pauseConn);
-        var pauseTask = pauseServer.ExecuteAsync("CLIENT", "PAUSE", 500);
+        var pauseTask = pauseServer.ExecuteAsync("CLIENT", "PAUSE", 2000);
 
         var key = Me();
         var db = conn.GetDatabase();