Fix flaky memberlist TestRejoin test (#598)

It randomly failed here: #525 and in some of the test runs here: #596

This test fails due to the probe interval being 5s, which is the same time we poll for

Author: julienduchesne

.github/workflows/find-flaky-tests.yml

@@ -8,6 +8,7 @@ jobs:
   find-flaky-tests:
     # Only run this workflow when the comment contains '/find-flaky-tests' or it's manually triggered
     if: contains(github.event.comment.body, '/find-flaky-tests') || github.event_name == 'workflow_dispatch'
+    fail-fast: false
     strategy:
       matrix:
         runs: [1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20] # Run the build workflow 20 times

kv/memberlist/memberlist_client_test.go

@@ -1240,20 +1240,20 @@ func TestRejoin(t *testing.T) {
 		return mkv2.memberlist.NumMembers()
 	}
 
-	poll(t, 5*time.Second, 2, membersFunc)
+	poll(t, 7*time.Second, 2, membersFunc) // Probe interval is 5s, with 2s timeout, so probe for 7s.
 
 	// Shutdown first KV
 	require.NoError(t, services.StopAndAwaitTerminated(context.Background(), mkv1))
 
 	// Second KV should see single member now.
-	poll(t, 5*time.Second, 1, membersFunc)
+	poll(t, 7*time.Second, 1, membersFunc)
 
 	// Let's start first KV again. It is not configured to join the cluster, but KV2 is rejoining.
 	mkv1 = NewKV(cfg1, log.NewNopLogger(), &dnsProviderMock{}, prometheus.NewPedanticRegistry())
 	require.NoError(t, services.StartAndAwaitRunning(context.Background(), mkv1))
 	defer services.StopAndAwaitTerminated(context.Background(), mkv1) //nolint:errcheck
 
-	poll(t, 5*time.Second, 2, membersFunc)
+	poll(t, 7*time.Second, 2, membersFunc)
 }
 
 func TestMessageBuffer(t *testing.T) {