Issue with Riva ASR Server Setup on Nvidia Jetson AGX Orin

Robotics & Edge Computing Jetson & Embedded Systems Jetson AGX Orin
,
1

Hello,

I am currently encountering an issue while setting up a Riva ASR server.

When I run the riva_start.sh script, I systematically receive the following message:
“Waiting for Riva server to load all models… Retrying in 10 seconds.” This message appears multiple times before ending with:

Health ready check failed.
Check Riva logs with: docker logs riva-speech.

Explanation:

I am working on an Nvidia Jetson AGX Orin (developer kit) equipped with an NVMe SSD. This machine is currently running JetPack 6.0 [L4T 36.3.0].

For the installation of the necessary components to use Riva, I followed these documentations:

During this last installation, I did not configure the config.sh file.

To resolve this issue, I tried accessing the logs with sudo docker logs riva-speech, but nothing is displayed.

While searching for solutions, I found that using the -x attribute provides more information during script execution. I am attaching the console output when running the riva_init and riva_start scripts with and without the -x.

Before posting here, I tried several solutions: reinstalling Docker, NGC, and Riva, changing permissions, but nothing worked.

Could someone please help me?

It is interesting to note that I do not see any significant errors in the console during the execution of the start and init scripts. However, I systematically get the following error when running riva_clean.sh:

bash riva_clean.sh
Cleaning up local Riva installation.
Image nvcr.io/nvidia/riva/riva-speech:2.19.0-l4t-aarch64 found. Delete? [y/N] y
Error response from daemon: get /home/uitml/ngc/riva_quickstart_arm64_v2.19.0/riva_quickstart_arm64_v2.19.0/model_repository: no such volume
'/home/uitml/ngc/riva_quickstart_arm64_v2.19.0/riva_quickstart_arm64_v2.19.0/model_repository' is not a Docker volume, or has already been deleted.
Found models at '/home/uitml/ngc/riva_quickstart_arm64_v2.19.0/riva_quickstart_arm64_v2.19.0/model_repository'. Delete? [y/N] y
[sudo] password for uitml:

Thank you in advance for your help.
init + start + -x.txt (32.7 KB)
init + start.txt (7.8 KB)

3

Hi,

Do you have any dependencies on JetPack 6.0?
If not, would you mind reflashing your system with the latest JetPack 6.2?

We have tested the 2.19.0 script on the latest software and it can work as expected.
Thanks.

4

Hi,

Yes, I need to work on JetPack 6.0 because TensorRT 10 (TRT 10) has removed the deprecated support for Caffe and TensorFlow UFF models.

However, if version 2.19 of the script does not work on JetPack 6.0, there might be other internal versions that would work, but I haven’t found any documentation discussing version compatibility.

If you can indicate which version of the script to install, I am willing to make the changes; I am not particularly attached to having the latest version of Riva ASR.

Thank you.

5

Hi,

Waiting for Riva server to load all models…retrying in 10 seconds

It turns out there are some issues in our server so the models cannot be downloaded.
We are checking this with our internal team and will provide more info to you later.

Thanks.

7

Hi,

Could you share the below command and share the output with us?

docker logs riva-speech

Thanks.

8

Hello,

As you suggested, I found a solution to switch to Jetpack 6.2. Therefore, I redid my entire installation on Jetpack 6.2.

However, I am attaching the console content again. This did not change anything about my problem.

I would like to clarify, in response to your last request, that when I try to run the command sudo docker logs riva-speech or docker logs riva-speech, it displays absolutely nothing. You can see this at the end of the document ‘init + start 2.txt’.

init + start + -x 2.txt (32.4 KB)
init + start 2.txt (8.1 KB)

9

Hi,

Thanks for your info.
We are checking with our internal team to gather more information.
Will get back to you later.

Thanks.

10

Hi,

Please find the below comment for the WAR:

Thanks.

12

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.