Ethernet Disconnects During Deep Stream RTSP Streaming on Jetson Orin Nano

Issue Overview

Users of the Nvidia Jetson Orin Nano Developer Kit are experiencing Ethernet disconnections when running a Deep Stream application that receives RTSP source input and performs RTSP streaming. The issue occurs after a certain period of time, and is accompanied by the message "r8168: eth0: link up" in the kern.log file. This problem affects the system’s network connectivity and interrupts the streaming process, impacting the overall functionality of the Deep Stream application.

Possible Causes

1. Driver issues: The r8168 Ethernet driver may have compatibility problems or bugs that cause intermittent disconnections.

2. Network configuration: Improper network settings (DHCP or static IP) might contribute to the instability of the Ethernet connection.

3. Deep Stream application load: The intensive processing required by the Deep Stream app could potentially interfere with network operations.

4. Hardware limitations: The Jetson Orin Nano’s hardware might struggle to handle the combined load of Deep Stream processing and network operations.

5. Software conflicts: Interactions between the Deep Stream application, the operating system, and network services could lead to unexpected behavior.

Troubleshooting Steps, Solutions & Fixes

1. Update JetPack:

   • Ensure you are using the latest version of JetPack. The issue was reported on JetPack 6.0, but newer versions may include fixes.

2. Apply r8168 driver patches:

   • NVIDIA has provided patches to upgrade the r8168 driver for rel-36 (JetPack 6.x). These patches are not yet included in the current release but will be in future versions.
   • To apply the patches manually:
     a. Download the provided patch files (928bbd8.diff, 954d58f.diff, 91baac0.diff, 6d4ef61.diff, 4643e35.diff).
     b. Apply the patches to the r8168 driver source code.
     c. Recompile and install the updated driver.

3. Network configuration:

   • Try both DHCP and static IP configurations to determine if the issue persists with both settings.
   • If using static IP, ensure all network parameters (IP address, subnet mask, gateway, DNS) are correctly configured.

4. Monitor system resources:

   • Use tools like top, htop, or nvidia-smi to monitor CPU, GPU, and memory usage during Deep Stream operation.
   • If resources are consistently maxed out, consider optimizing the Deep Stream application or reducing the workload.

5. Check for thermal throttling:

   • Monitor system temperature using tegrastats or other thermal monitoring tools.
   • Ensure proper cooling for the Jetson Orin Nano, especially during intensive Deep Stream operations.

6. Analyze log files:

   • Regularly check kern.log and journalctl logs for any error messages or warnings related to the Ethernet interface or Deep Stream application.
   • Look for patterns or specific events that occur just before the disconnections.

7. Test with a simplified Deep Stream configuration:

   • Create a minimal Deep Stream application configuration that still reproduces the issue.
   • Gradually add complexity to identify which specific settings or operations trigger the problem.

8. Ethernet interface settings:

   • Try adjusting Ethernet interface settings such as speed, duplex mode, or power management features:

     sudo ethtool -s eth0 speed 100 duplex full autoneg off
     

   • Disable power management for the Ethernet interface:

     sudo ethtool -s eth0 wol d
     

9. Kernel parameter adjustments:

   • Modify kernel parameters related to network behavior. For example, to increase the netdev backlog queue:

     sudo sysctl -w net.core.netdev_max_backlog=5000
     

   • Make changes permanent by adding them to /etc/sysctl.conf.

10. Wait for future updates:

    • As NVIDIA has acknowledged the issue and prepared patches, future releases of JetPack (after 6.0) should include fixes for the r8168 driver.
    • Regularly check for JetPack updates and apply them when available.

If the issue persists after trying these solutions, consider reaching out to NVIDIA support with detailed logs and a minimal reproducible example of your Deep Stream application configuration.