Memory Leak Issue with NvArgus on Jetson Orin Nano

Issue Overview

Users are experiencing a memory leak issue with the NvArgus daemon on the Jetson Orin Nano development board, specifically when using applications written with Argus. The symptoms include the daemon stopping due to event overflow after processing approximately 64,975 frames. This issue has been reported on JetPack 5.1.1, with users seeking alternatives to upgrading to JetPack 5.1.2 to resolve the problem.

The problem occurs during the execution of Argus applications, and users have noted that it does not seem to affect command-line operations using GStreamer. The memory leak can be monitored using tools like JTOP, which indicates increasing memory consumption until the application fails. The impact of this issue is significant, as it disrupts application performance and can lead to system instability.

Possible Causes

• Software Bugs or Conflicts: The issue may stem from bugs within the Argus framework or conflicts with other software components, particularly related to memory management.

• Driver Issues: Problems with the underlying drivers for the Jetson Orin Nano could contribute to improper memory handling, especially concerning the NvJpegEncoder, which is utilized in Argus applications.

• Configuration Errors: Incorrect configuration settings within Argus applications might lead to inefficient memory usage or leaks.

• Environmental Factors: Factors such as power supply inconsistencies or temperature fluctuations could exacerbate memory management issues.

• User Errors or Misconfigurations: Users may inadvertently misconfigure their applications, leading to increased memory usage or leaks.

Troubleshooting Steps, Solutions & Fixes

1. Diagnosing the Problem:

   • Use JTOP to monitor memory usage while running your Argus application.
   • Check logs for any error messages related to memory allocation or daemon failures.

2. Gathering System Information:

   • Execute the following command to check for available memory and processes:

     free -h
     top
     

3. Isolating the Issue:

   • Test your application with different configurations or parameters.
   • Run sample applications from the jetson_multimedia_api, such as the 09 argus jpeg sample, to replicate the issue and confirm it’s not specific to your code.

4. Potential Fixes and Workarounds:

   • If upgrading to JetPack 5.1.2 is not an option, consider switching to a software JPEG encoder as a temporary solution.
   • Review and optimize your Argus application code for better memory management practices.
   • Ensure that all dependencies are up-to-date and compatible with your version of JetPack.

5. Documentation and Updates:

   • Keep an eye on NVIDIA’s forums and documentation for any patches or updates related to Argus and JetPack.
   • Check for driver updates that may address known issues with memory leaks.

6. Best Practices for Prevention:

   • Regularly monitor your application’s performance and memory usage during development.
   • Implement logging within your application to capture detailed information about memory allocation patterns.

Unresolved Aspects

While several users have reported success with using software encoders as a workaround, further investigation is needed into whether this issue will be addressed in future JetPack updates or if there are deeper underlying problems with the Argus framework itself.