Deepstream app slows if any of the rtsp sources become flaky

[arivarasan.e]

• Hardware Platform (Jetson / GPU) - GPU
• DeepStream Version - 6.1
• JetPack Version (valid for Jetson only)
• TensorRT Version - 8.4.1
• NVIDIA GPU Driver Version (valid for GPU only) - 520.61.05
• Issue Type( questions, new requirements, bugs) - question

My deepstream app runs fine if I run with one single rtsp source. However if I run 8 rtsp sources, the fps drops very significantly, sometimes becomes 0 with no processing at all. My question is, if any of the rtsp sources is flaky, does it impact fps of the other streams as well? If so, is there anyway to detect the flakiness of the rtsp source? So that, the corresponding source can be swapped with videotestsrc to not to impact fps on other streams.
Note: I already have a runtime addition/deletion of source logic in place to swap the rtsp source with videotestsrc in case the source sends EOS. But the problem is, the source doesn’t send EOS, it’s just flaky in nature.

[arivarasan.e]

Team, Any clarification on this please?

[yuweiw]

You can refer the link below and try to set the batched-push-timeout paras.
https://docs.nvidia.com/metropolis/deepstream/dev-guide/text/DS_plugin_gst-nvstreammux.html

[arivarasan.e]

I’m using the new nvstreammux. Is batched-push-timeout still valid with new nvstreammux?

[yuweiw]

There is no update from you for a period, assuming this is not an issue anymore. Hence we are closing this topic. If need further support, please open a new one. Thanks

Yes, you can refer new nvstreammux: https://docs.nvidia.com/metropolis/deepstream/dev-guide/text/DS_plugin_gst-nvstreammux2.html

[arivarasan.e]

This helps, thanks!