HPROF dump file

Tags

Creating a Java HPROF Dump

An HPROF (Heap Profiler) dump is a snapshot of a Java application’s heap memory, which is useful for diagnosing memory leaks, analyzing memory usage, and debugging performance issues.

There are several ways to generate an HPROF dump in Java:

1. Using jmap Command-Line Tool

jmap is a JDK tool that allows you to obtain heap dumps of a running Java process.

Steps:

  1. Find the process ID (PID) of the Java application: 
    jps

    Example output:

    12345 MyJavaApplication
  2. Generate the heap dump using jmap: 
    jmap -dump:format=b,file=heapdump.hprof 12345
    • format=b → Binary format
    • file=heapdump.hprof → Output file
    • 12345 → Java process ID

Example:

jmap -dump:format=b,file=heap.hprof 9876

Location: The dump file (heap.hprof) will be created in the working directory.

2. Using jcmd Command

jcmd is a more advanced alternative to jmap.

Steps:

  1. Find the Java process ID: 
    jps
  2. Run the following command: 
    jcmd <PID> GC.heap_dump heapdump.hprof

    Example:

    jcmd 9876 GC.heap_dump heapdump.hprof

Advantages of jcmd:

  • Works on newer Java versions (jmap might not work on some JVM versions)
  • Safer than jmap for heap dumps

3. Using JVM Startup Options (-XX:+HeapDumpOnOutOfMemoryError)

If you want an automatic heap dump when an OutOfMemoryError occurs:

  1. Start your Java application with: 
    java -XX:+HeapDumpOnOutOfMemoryError -XX:HeapDumpPath=/path/to/dump MyJavaApplication
    • -XX:+HeapDumpOnOutOfMemoryError → Generates a heap dump when an OOM error happens.
    • -XX:HeapDumpPath=/path/to/dump → Specifies the dump file path.

4. Programmatically Trigger Heap Dump in Java

If you want to trigger an HPROF dump from inside a Java program, you can use:

Using com.sun.management.HotSpotDiagnostic

import com.sun.management.HotSpotDiagnostic;
import java.lang.management.ManagementFactory;
import java.io.IOException;

public class HeapDumpGenerator {
    public static void dumpHeap(String filePath, boolean live) throws IOException {
        HotSpotDiagnostic mxBean = ManagementFactory.getPlatformMXBean(HotSpotDiagnostic.class);
        mxBean.dumpHeap(filePath, live);
    }

    public static void main(String[] args) {
        try {
            dumpHeap("heapdump.hprof", true);
            System.out.println("Heap dump created.");
        } catch (IOException e) {
            e.printStackTrace();
        }
    }
}
  • dumpHeap("heapdump.hprof", true); → Dumps only live objects (true) or the full heap (false).

5. Using kill -3 (For UNIX/Linux)

  • Send a SIGQUIT (kill -3) signal to the Java process: 
    kill -3 <PID>
  • This prints a heap dump into the Java application’s standard output (e.g., catalina.out for Tomcat).

Where is the HPROF Dump Stored?

  • By default, the .hprof file is stored in the Java application’s working directory unless a specific path is given.
  • If using HeapDumpOnOutOfMemoryError, it will be stored in the same directory as the running Java process unless -XX:HeapDumpPath=<path> is set.

How to Analyze an HPROF File?

Once you have the heapdump.hprof file, you can analyze it using:

  • Eclipse Memory Analyzer (MAT): 
    mat heapdump.hprof
  • VisualVM:
    Open VisualVM (jvisualvm), load the .hprof file and inspect the heap.
  • JProfiler (Commercial)

Comparison of Methods

| Method | Requires PID | Triggers Heap Dump | Notes | |——–|————-|——————-|——-| | jmap | ✅ | ✅ | May not work in newer Java versions | | jcmd | ✅ | ✅ | Preferred for Java 8+ | | -XX:+HeapDumpOnOutOfMemoryError | ❌ | Auto | Dumps on OOM automatically | | kill -3 | ✅ | ❌ | Outputs heap info to logs | | Java Code | ❌ | ✅ | Can be triggered programmatically |

Conclusion

  • If you’re debugging a running process, use jcmd <PID> GC.heap_dump.
  • If you want a dump on OOM, use -XX:+HeapDumpOnOutOfMemoryError.
  • If you need to programmatically generate a heap dump, use the HotSpotDiagnostic API.
  • If jmap fails, try jcmd, which is more reliable on newer Java versions.

Would you like help analyzing an HPROF file? 🚀

LinkedIn Reddit