[NVIDIA VSS] Exception: Failed to generate TRT-LLM engine

Accelerated Computing Intelligent Video Analytics Visual AI Agent
1

System & Server Information

  • Cloud Provider: JarvisLabs
  • Hardware: 8 x H100 (80 GB)
  • System Memory: 200 GB
  • OS: Ubuntu 22.04
  • NVIDIA GPU Driver Version: 535.216.03
  • Issue Type: Bug

Description of the Issue
I am attempting to deploy the NVIDIA Video Search and Summarization Agent (VSS) using Microk8s, following the official quickstart guide without any additional configuration or override values.

Previously, on February 12, 2025, I successfully installed NVIDIA VSS using Helm chart version 2.1.0. However, when attempting the same installation on March 04, 2025, I encountered errors with both Helm chart versions 2.1.0 and 2.2.0. The errors are consistent across both versions.

Steps to Reproduce

  1. Follow the official NVIDIA VSS quickstart guide.
  2. Use Helm chart versions 2.1.0 or 2.2.0 without additional configuration.
  3. Deploy the VSS.

NVIDIA VSS - 2.2.0.txt (6.0 KB)
NVIDIA VSS - 2.1.0.txt (4.7 KB)

3

Have you remove any stale TensorRT engines for VILA-1.5?

5

Yes, I have. During the initial installation, the server was clean (fresh), and I installed version 2.1.0. Later, after checking the documentation, I noticed an upgrade to version 2.2.0. I uninstalled the Helm chart, removed the PVCs, and installed the new version, but the issue persisted. To confirm, I also checked the host path, and it was already cleared.

6

OK. Then could you refer to this https://forums.developer.nvidia.com/t/vss-blueprint-2-2-0-error-failed-to-load-via-stream-handler-failed-to-generate-trt-llm-engine/325745/7 and see if it can fix your issue?

7

I wonder if this fix can be pushed to the helm chart?

8

Just from your log attached, ValueError: Unrecognized model in /tmp/tmp.vila.J4Wd8KGk.. It might be that the model is not properly downloaded, so you can try it.

9

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

10

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