How to Diagnose 100% CPU Spikes in Java Production Servers (Windows & Linux)

Introduction

Production servers have several Java programs, suddenly a CPU 100% alert appears; how do you locate the problem?

The article presents two answer styles and then focuses on the practical troubleshooting steps.

Body

The same three‑step procedure applies to both Windows and Linux environments, with only the commands differing.

Find the PID of the process consuming the most CPU.

Find the thread ID (TID) within that PID that consumes the most CPU.

Locate the corresponding Java thread and address the issue.

Windows version

Many government systems run on Windows Server, so the Windows workflow is described.

Find the PID

Using the Windows Task Manager (or similar) to display the PID column, the highest‑CPU process is identified as PID 10856 .

Find the thread ID

Process Explorer (v16.22) is used to inspect threads. The most CPU‑intensive thread has TID 6616 , which is 0x19d8 in hexadecimal.

Locate the Java thread

Export a process dump, open the file c:/10856.stack, and search for 0x19d8. The stack trace shows that TestFor.java line 7 is looping, pinpointing the problem.

Linux version

Find the PID

Run top -c and press P to sort by CPU usage. The process with PID 3033 uses the most CPU.

Find the thread ID

Run top -Hp 3033 to list threads of that process, sort by CPU, and identify thread PID 3034 as the highest consumer. Its hexadecimal ID is 0xbda.

Locate the Java thread

Export a stack dump with: jstack -l 3033 > ./3033.stack Then search the dump:

cat 3033.stack | grep 'bda' -C 8

The output reveals the problematic Java code, completing the diagnosis.

Conclusion

Remember to actually perform the steps on a test environment to verify the fix.