[rllib] Add copy() in async samples optimizer to fix memory leak (#3938)

Fixes #3884.
ray-project · Feb 4, 2019 · 7ef830b · 7ef830b
1 parent 8323419
commit 7ef830b
Showing 1 changed file with 2 additions and 1 deletion.
diff --git a/python/ray/rllib/optimizers/async_samples_optimizer.py b/python/ray/rllib/optimizers/async_samples_optimizer.py
@@ -167,7 +167,8 @@ def _step(self):
                    for b in self.batch_buffer) >= self.train_batch_size:
                 train_batch = self.batch_buffer[0].concat_samples(
                     self.batch_buffer)
-                self.learner.inqueue.put(train_batch)
+                # defensive copy against plasma ref count bugs, see #3884
+                self.learner.inqueue.put(train_batch.copy())
                 self.batch_buffer = []
 
             # If the batch was replayed, skip the update below.