• Hardware Platform (Jetson / GPU)
RTX 4090
• DeepStream Version
6.3
• TensorRT Version
8.5.3-1
• NVIDIA GPU Driver Version (valid for GPU only)
530.30.02
• Issue Type( questions, new requirements, bugs)
This is a follow-up for the topic Custom nvmsgconv gst_element_factory_make no such element "nvmsgconv" .
The issue explained in above topic persists after upgrading to Deepstream 6.3, currently I am using the docker image nvcr.io/nvidia/deepstream:6.3-gc-triton-devel .
I also re-wrote my version of the nvmsgconv-library. Now I am using the following config for the msgconv element:
mdgconv:
payload-type: 0
msg2-newapi: 1
And I access the frame meta I want to populate the message with in dsmeta_payload.cpp in the nvmsgconv-library.
There is some problem connected to the nvmsgconv-library causing this issue since the msgconv-element is created correctly when using the original version of the library, but I don’t understand what might be causing the issue ( since the errors occurs already when trying to create the nvmsgconv-element) and why it only occurs sometimes. I even got it to work a few times on the machine that it previously did not work, only to fail again at a later moment. I could really use some insight as of how to debug this.
Any ideas of what could be the cause of no such element factory "nvmsgconv"? Is it connected to any specific part in the nvsmgconv-library? Could it be connected to something else, for example the kafka broker outside of the deepstream app? Could there be something wrong with the linking of libraries when building the nvmsgconv library, such that the library containing the needed factory method could not be found?