Work around JVM Bug in LongGCDisruptionTests (#50731) (#50974)

There is a JVM bug causing `Thread#suspend` calls to randomly take
multiple seconds breaking these tests that call the method numerous times
in a loop. Increasing the timeout would will not work since we may call
`suspend` tens if not hundreds of times and even a small number of them
experiencing the blocking will lead to multiple minutes of waiting.

This PR detects the specific issue by timing the `Thread#suspend` calls and
skips the remainder of the test if it timed out because of the JVM bug.

Closes #50047

test/framework/src/main/java/org/elasticsearch/test/disruption/LongGCDisruption.java

@@ -32,6 +32,8 @@ import java.util.Arrays;
 import java.util.Random;
 import java.util.Set;
 import java.util.concurrent.ConcurrentHashMap;
+import java.util.concurrent.TimeUnit;
+import java.util.concurrent.atomic.AtomicBoolean;
 import java.util.concurrent.atomic.AtomicReference;
 import java.util.regex.Pattern;
 import java.util.stream.Collectors;
@@ -56,11 +58,23 @@ public class LongGCDisruption extends SingleNodeDisruption {
     private Set<Thread> suspendedThreads;
     private Thread blockDetectionThread;
 
+    private final AtomicBoolean sawSlowSuspendBug = new AtomicBoolean(false);
+
     public LongGCDisruption(Random random, String disruptedNode) {
         super(random);
         this.disruptedNode = disruptedNode;
     }
 
+    /**
+     * Checks if during disruption we ran into a known JVM issue that makes {@link Thread#suspend()} calls block for multiple seconds
+     * was observed.
+     * @see <a href=https://bugs.openjdk.java.net/browse/JDK-8218446>JDK-8218446</a>
+     * @return true if during thread suspending a call to {@link Thread#suspend()} took more than 3s
+     */
+    public boolean sawSlowSuspendBug() {
+        return sawSlowSuspendBug.get();
+    }
+
     @Override
     public synchronized void startDisrupting() {
         if (suspendedThreads == null) {
@@ -251,7 +265,11 @@ public class LongGCDisruption extends SingleNodeDisruption {
                          * assuming that it is safe.
                          */
                         boolean definitelySafe = true;
+                        final long startTime = System.nanoTime();
                         thread.suspend();
+                        if (System.nanoTime() - startTime > TimeUnit.SECONDS.toNanos(3L)) {
+                            sawSlowSuspendBug.set(true);
+                        }
                         // double check the thread is not in a shared resource like logging; if so, let it go and come back
                         safe:
                         for (StackTraceElement stackElement : thread.getStackTrace()) {

test/framework/src/test/java/org/elasticsearch/test/disruption/LongGCDisruptionTests.java

@@ -18,7 +18,6 @@
  */
 package org.elasticsearch.test.disruption;
 
-import org.elasticsearch.bootstrap.JavaVersion;
 import org.elasticsearch.common.Nullable;
 import org.elasticsearch.test.ESTestCase;
 
@@ -115,8 +114,6 @@ public class LongGCDisruptionTests extends ESTestCase {
      * but does keep retrying until all threads can be safely paused
      */
     public void testNotBlockingUnsafeStackTraces() throws Exception {
-        assumeFalse("https://github.com/elastic/elasticsearch/issues/50047",
-            JavaVersion.current().equals(JavaVersion.parse("11")) || JavaVersion.current().equals(JavaVersion.parse("12")));
         final String nodeName = "test_node";
         LongGCDisruption disruption = new LongGCDisruption(random(), nodeName) {
             @Override
@@ -149,7 +146,14 @@ public class LongGCDisruptionTests extends ESTestCase {
                 threads[i].start();
             }
             // make sure some threads are under lock
-            disruption.startDisrupting();
+            try {
+                disruption.startDisrupting();
+            } catch (RuntimeException e) {
+                if (e.getMessage().contains("suspending node threads took too long") && disruption.sawSlowSuspendBug()) {
+                    return;
+                }
+                throw new AssertionError(e);
+            }
             long first = ops.get();
             assertThat(lockedExecutor.lock.isLocked(), equalTo(false)); // no threads should own the lock
             Thread.sleep(100);
@@ -157,6 +161,7 @@ public class LongGCDisruptionTests extends ESTestCase {
             disruption.stopDisrupting();
             assertBusy(() -> assertThat(ops.get(), greaterThan(first)));
         } finally {
+            disruption.stopDisrupting();
             stop.set(true);
             for (final Thread thread : threads) {
                 thread.join();