Please provide complete information as applicable to your setup.
• Hardware Platform (Jetson Nano Development Kit)
• DeepStream Version 5.0 GA
• JetPack Version (4.4)
• TensorRT Version
We installed the latest JetPack (on a fresh SD card) that is compatible with Deepstream 5.0 GA and followed the instructions provided in DeepStream SDK 5.0 GA Getting Started guide for Jetson.
When we have one RTSP input stream (of 1080p, 30FPS coming from a VLC runing on Jetson) and One RTSP output, with just Primary Inference + Tracker + OSD in the pipeline. This results in some kind of network storm and the response from a terminal connected via SSH on Jetson becomes very very slow. The VLC client running on a PC is unable to connect and view the RTSP output.
The application works well when used with Egl displayed on the local screen.
The above RTSP-in RTSP-out configuration had worked well with DS 5.0 DP version (and corresponding JetPack).
Has anyone seen this issues? Is there any special network configuration required for this?
This is a bottleneck and we are unable to make further analysis/progress on DS 5.0 GA.
Any help will be appreciated.