How to Detect and Fix Java Native Memory Leaks: Tools, Symptoms, and Solutions

What Is Java Native Memory?

Java native memory resides outside the heap and includes areas such as metaspace, thread stacks, code cache, direct buffers, GC internal structures, JNI allocations, and other small but critical JVM components. Unlike heap memory, it is not managed by the JVM garbage collector and is controlled by the operating system or native code.

Key Native Memory Regions in the JVM

Metaspace : Stores class metadata; grows on demand and can be limited with -XX:MaxMetaspaceSize . Exceeds limit triggers java.lang.OutOfMemoryError: Metaspace .

Threads : Each thread’s stack and internal structures consume native memory; stack size set with -Xss . Too many threads cause java.lang.OutOfMemoryError: unable to create new native thread .

Code Cache : Holds compiled native code; when full, JVM throws java.lang.OutOfMemoryError: CodeCache is full . Size adjustable via -XX:ReservedCodeCacheSize .

Direct Buffers : Off‑heap memory allocated via ByteBuffer.allocateDirect(); size limited by -XX:MaxDirectMemorySize . Exhaustion leads to java.lang.OutOfMemoryError: Direct buffer memory .

Garbage Collector (GC) : Uses native structures (mark‑bitmap, card tables) that affect overall native memory usage; monitoring often requires enabling Native Memory Tracking.

JNI : Native code allocations (malloc, calloc) are outside JVM control and must be freed manually.

Other : Miscellaneous JVM internal allocations that can indicate pressure when they grow.

Symptoms of Native Memory Leaks

Heap memory usage appears stable while the process’s overall memory steadily increases.

GC runs frequently but cannot reclaim memory, yet heap metrics remain normal.

How to Detect Which Native Memory Region Is Leaking

1. Native Memory Tracking (NMT)

Enable NMT when starting the JVM:

java -XX:NativeMemoryTracking=summary -jar YourApplication.jar

Periodically capture memory summaries (e.g., every 10 minutes):

jcmd <pid> VM.native_memory summary > nmt_report.txt

Upload the report to GCeasy for visual analysis of region growth.

2. OutOfMemoryError Types

java.lang.OutOfMemoryError: Metaspace – Affects Metaspace.

java.lang.OutOfMemoryError: unable to create new native thread – Affects Thread stacks.

java.lang.OutOfMemoryError: Direct buffer memory – Affects Direct Buffers.

java.lang.OutOfMemoryError: Code Cache full – Affects Code Cache.

Common Native Memory Leak Scenarios

1. Metaspace Leak

Occurs when dynamically loaded classes (e.g., in Spring, Hibernate, OSGi) are not unloaded, causing metaspace to grow until the Metaspace OOM error is thrown.

2. Thread Leak

Each thread consumes native stack memory; creating threads without proper shutdown eventually triggers java.lang.OutOfMemoryError: unable to create new native thread . Causes include buggy code, insufficient RAM, excessive concurrent processes, or kernel thread limits.

3. Direct Buffer Leak

Direct buffers are allocated off‑heap and rely on explicit release. Leaks happen due to code defects, high allocation rates, or frameworks (e.g., switching from RestTemplate to WebClient) that shift objects to direct buffers.

Conclusion

Although Java native memory leaks are rare, they can silently degrade applications. By understanding JVM native memory regions, monitoring symptoms, and using tools like Native Memory Tracking and OutOfMemoryError analysis, developers can detect and resolve leaks before they cause production failures.