From b2456b7b104b92007e6d937851973ca7c95f1479 Mon Sep 17 00:00:00 2001
From: Eduard Tudenhoefner <etudenhoefner@gmail.com>
Date: Fri, 19 Jun 2026 15:06:46 +0200
Subject: [PATCH 1/2] GH-3558: Properly close buffers for vectored IO

---
 .../parquet/hadoop/ParquetFileReader.java     | 29 +++++++++++++++++--
 pom.xml                                       |  2 +-
 2 files changed, 28 insertions(+), 3 deletions(-)

diff --git a/parquet-hadoop/src/main/java/org/apache/parquet/hadoop/ParquetFileReader.java b/parquet-hadoop/src/main/java/org/apache/parquet/hadoop/ParquetFileReader.java
index e0b0d76e0e..95914bcb2a 100644
--- a/parquet-hadoop/src/main/java/org/apache/parquet/hadoop/ParquetFileReader.java
+++ b/parquet-hadoop/src/main/java/org/apache/parquet/hadoop/ParquetFileReader.java
@@ -65,6 +65,7 @@
 import org.apache.parquet.HadoopReadOptions;
 import org.apache.parquet.ParquetReadOptions;
 import org.apache.parquet.Preconditions;
+import org.apache.parquet.bytes.ByteBufferAllocator;
 import org.apache.parquet.bytes.ByteBufferInputStream;
 import org.apache.parquet.bytes.ByteBufferReleaser;
 import org.apache.parquet.bytes.BytesInput;
@@ -1361,8 +1362,32 @@ private void readVectored(List<ConsecutivePartList> allParts, ChunkListBuilder b
       totalSize += len;
     }
     LOG.debug("Reading {} bytes of data with vectored IO in {} ranges", totalSize, ranges.size());
-    // Request a vectored read;
-    f.readVectored(ranges, options.getAllocator());
+    // Wrap the allocator to track buffers allocated during the vectored read.
+    // Hadoop's vectored IO may merge ranges and return slices of merged buffers,
+    // so the buffers returned via CompletableFuture may differ from those originally
+    // allocated. We track the originals here to ensure they are properly released.
+    ByteBufferAllocator baseAllocator = options.getAllocator();
+    List<ByteBuffer> allocatedBuffers = new ArrayList<>();
+    ByteBufferAllocator trackingAllocator = new ByteBufferAllocator() {
+      @Override
+      public ByteBuffer allocate(int size) {
+        ByteBuffer buf = baseAllocator.allocate(size);
+        allocatedBuffers.add(buf);
+        return buf;
+      }
+
+      @Override
+      public void release(ByteBuffer b) {
+        baseAllocator.release(b);
+      }
+
+      @Override
+      public boolean isDirect() {
+        return baseAllocator.isDirect();
+      }
+    };
+    f.readVectored(ranges, trackingAllocator);
+    builder.addBuffersToRelease(allocatedBuffers);
     int k = 0;
     for (ConsecutivePartList consecutivePart : allParts) {
       ParquetFileRange currRange = ranges.get(k++);
diff --git a/pom.xml b/pom.xml
index a0a6678163..84b61fe81e 100644
--- a/pom.xml
+++ b/pom.xml
@@ -83,7 +83,7 @@
     <spotless.version>2.46.1</spotless.version>
     <shade.prefix>shaded.parquet</shade.prefix>
     <!-- Guarantees no newer classes/methods/constants are used by parquet. -->
-    <hadoop.version>3.3.0</hadoop.version>
+    <hadoop.version>3.5.0</hadoop.version>
     <parquet.format.version>2.13.0</parquet.format.version>
     <previous.version>1.17.0</previous.version>
     <thrift.executable>thrift</thrift.executable>

From 676f1be06e436a80b1386f76d30d9e1ebc0f1363 Mon Sep 17 00:00:00 2001
From: Eduard Tudenhoefner <etudenhoefner@gmail.com>
Date: Fri, 19 Jun 2026 15:18:05 +0200
Subject: [PATCH 2/2] Update pom.xml

---
 pom.xml | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/pom.xml b/pom.xml
index 84b61fe81e..45ad9118e4 100644
--- a/pom.xml
+++ b/pom.xml
@@ -83,7 +83,7 @@
     <spotless.version>2.46.1</spotless.version>
     <shade.prefix>shaded.parquet</shade.prefix>
     <!-- Guarantees no newer classes/methods/constants are used by parquet. -->
-    <hadoop.version>3.5.0</hadoop.version>
+    <hadoop.version>3.4.3</hadoop.version>
     <parquet.format.version>2.13.0</parquet.format.version>
     <previous.version>1.17.0</previous.version>
     <thrift.executable>thrift</thrift.executable>