use final atomicreference for warmed compiled execution to avoid sync…

…hronization

logstash-core/src/main/java/org/logstash/config/ir/CompiledPipeline.java

@@ -36,6 +36,7 @@
 import java.util.Map;
 import java.util.Objects;
 import java.util.concurrent.ConcurrentHashMap;
+import java.util.concurrent.atomic.AtomicReference;
 import java.util.stream.Collectors;
 import java.util.stream.Stream;
 
@@ -90,7 +91,7 @@ public final class CompiledPipeline {
      * the {@link CompiledPipeline#datasetClassCache} in a thread safe way
      * before the concurrent per worker threads {@link CompiledExecution} compilations
      */
-    private CompiledExecution warmedCompiledExecution;
+    private final AtomicReference<CompiledExecution> warmedCompiledExecution = new AtomicReference<>();
 
     public CompiledPipeline(
             final PipelineIR pipelineIR,
@@ -113,7 +114,7 @@ public CompiledPipeline(
 
             // invoke a first compilation to warm the class cache which will prevent
             // redundant compilations for each subsequent worker {@link CompiledExecution}
-            warmedCompiledExecution = new CompiledPipeline.CompiledExecution();
+            warmedCompiledExecution.set(new CompiledPipeline.CompiledExecution());
         } catch (Exception e) {
             throw new IllegalStateException("Unable to configure plugins: " + e.getMessage());
         }
@@ -137,14 +138,9 @@ public Collection<IRubyObject> inputs() {
      * @return Compiled {@link Dataset} representation of the underlying {@link PipelineIR} topology
      */
     public Dataset buildExecution() {
-        synchronized (this) {
-            // the first worker get the warmed CompiledExecution and the other
-            // get their own.
-            if (warmedCompiledExecution != null) {
-                final CompiledExecution result = warmedCompiledExecution;
-                warmedCompiledExecution = null;
-                return result.toDataset();
-            }
+        CompiledExecution result = warmedCompiledExecution.getAndSet(null);
+        if(result != null) {
+            return result.toDataset();
         }
         return new CompiledPipeline.CompiledExecution().toDataset();
     }