“Unable to set device in gst_nvstreammux_change_state” is a new error. it is related to CUDA.
please make the cmd was run in path of /opt/nvidia/deepstream/deepstream/samples/configs/deepstream-app-triton.
are you testing in DS7.0 nvcr.io/nvidia/deepstream:7.0-samples-multiarch docker? I can’t reproduce this issue in the docker container. Here is my test log log-0528.txt (8.1 KB). if you are not testing in docker, please make all components in this link are installed. then please make sure the simplest sample deepstream-test1 can run well.
I am inside container. I used this command from quickstart DS documention:
docker pull nvcr.io/nvidia/deepstream:7.0-triton-multiarch
I am getting csv error when running config file according to README in deepstream-test1 and same "gst_nvstreammux_change_state” error when running the c file
Also, the readme says for nvinfer we have to be inside inside DS triton docker exactly what I am using right now. Or despite being inside triton docker container, I have do set some other parameters?
if using docker, please refer to the doc. please make sure nvidia-docker and NVIDIA GPU Driver are installed successfully and use the steps in the " Run the container" to start the docker.
deepstream-test1 and deepstream-app are the different applications. if testing deepstream-test1, please refer to readme in /opt/nvidia/deepstream/deepstream-7.0/sources/apps/sample_apps/deepstream-test1. here are the steps.
1> export CUDA_VER=12.2
2> open deepstream_test1_app.c , then replace nveglglessink with fakesink, then make.
3> execute ./deepstream-test1-app dstest1_config.yml
Hi
I do not have x11 forwarding enabled as I am on VPN to a university’s network.
Is there any way to skip this step?
I have been able to run few deepstream apps in past without this option.
Also, could you please tell me how ./ is working to run yml as a script?
I can only get the following error message after making fakesink change using
deepstream-app -c
please refer to my last comment, you need to execute “make” to build the deepstream-test1-app, then execute ./deepstream-test1-app dstest1_config.yml. please refer to the readme for more details.
deepstream-app can’t use dstest1_config.yml, which is for deeptream-app. the yml files for depstream-app are in /opt/nvidia/deepstream/deepstream/samples/configs.
Th cuda error was because NVIDIA-SMI was not working in that docker container. I set up a new container with the above steps and app run was successful. Thanks!
