fix: race condition in downing provider spec (#31213) #31194

akka-cluster/src/test/scala/akka/cluster/DowningProviderSpec.scala

@@ -5,13 +5,10 @@
 package akka.cluster
 
 import java.util.concurrent.atomic.AtomicBoolean
-
 import scala.concurrent.duration._
-
 import com.typesafe.config.ConfigFactory
 import org.scalatest.matchers.should.Matchers
 import org.scalatest.wordspec.AnyWordSpec
-
 import akka.ConfigurationException
 import akka.actor.ActorSystem
 import akka.actor.Props
@@ -19,6 +16,8 @@ import akka.testkit.TestKit.awaitCond
 import akka.testkit.TestKit.shutdownActorSystem
 import akka.util.unused
 
+import scala.util.control.NonFatal
+
 class FailingDowningProvider(@unused system: ActorSystem) extends DowningProvider {
   override val downRemovalMargin: FiniteDuration = 20.seconds
   override def downingActorProps: Option[Props] = {
@@ -76,18 +75,30 @@ class DowningProviderSpec extends AnyWordSpec with Matchers {
     }
 
     "stop the cluster if the downing provider throws exception in props method" in {
-      val system = ActorSystem(
+      // race condition where the downing provider failure can be detected and trigger
+      // graceful shutdown fast enough that creating the actor system throws on constructing
+      // thread (or slow enough that we have time to try join the cluster before noticing)
+      val maybeSystem = try {
+        Some(
+          ActorSystem(
             "auto-downing",
             ConfigFactory.parseString("""
           akka.cluster.downing-provider-class="akka.cluster.FailingDowningProvider"
-        """).withFallback(baseConf))
+        """).withFallback(baseConf)))
+      } catch {
+        case NonFatal(_) =>
+          // expected to sometimes happen
+          None
+      }
 
+      maybeSystem.foreach { system =>
         val cluster = Cluster(system)
         cluster.join(cluster.selfAddress)
 
         awaitCond(cluster.isTerminated, 3.seconds)
         shutdownActorSystem(system)
       }
+    }
 
   }
 }