Can't install h5py nor libhdf5-dev

Robotics & Edge Computing Jetson & Embedded Systems Jetson Orin Nano
1

Hi all,

I am new to working on this and I am trying to get a docker build to work. Here is what I believe are the relevant parts of the dockerfile.

FROM nvcr.io/nvidia/l4t-tensorrt:r8.4.1.5-devel

ARG HOME=/app
ENV DEBIAN_FRONTEND noninteractive

RUN apt-get update && \
    apt-get install --no-install-recommends -y tcl build-essential gcc ffmpeg libsm6 libxext6 python3-dev && \
    apt-get clean && rm -rf /var/lib/apt/lists/*

#Makefile occurs here

RUN /usr/local/bin/pip3 install --no-cache-dir -r req.txt

In my req.txt I have tensorflow==2.13.1. This is resulting in this error:

#0 43.45 error: Unable to load dependency HDF5, make sure HDF5 is installed properly #0 43.45 Library dirs checked: [] #0 43.45 error: libhdf5.so: cannot open shared object file: No such file or directory #0 43.45 [end of output]

Everything I have looked up online indicates that this is missing libhdf5. I have tried adding libhdf5-dev and libhdf5-serial-dev to my apt-get install and they all end up with this error:

0 11.42 Err:13 http://ports.ubuntu.com/ubuntu-ports focal-updates/main arm64 libjpeg-turbo8-dev arm64 2.0.3-0ubuntu1.20.04.1
#0 11.42   404  Not Found [IP: 185.125.190.39 80]
#0 11.43 Fetched 6468 kB in 3s (2379 kB/s)
#0 11.43 E: Failed to fetch http://ports.ubuntu.com/ubuntu-ports/pool/main/libj/libjpeg-turbo/libjpeg-turbo8-dev_2.0.3-0ubuntu1.20.04.1_arm64.deb  404  Not Found [IP: 185.125.190.39 80]
#0 11.43 E: Unable to fetch some archives, maybe run apt-get update or try with --fix-missing?

Is this a problem that I am using the wrong tensorrt version?

3

Hi,

Which JetPack do you use?
The container you used is for the JetPack 5.0.2 environment.

Thanks.

4

How do I find that? I tried

sudo apt-cache show nvidia-jetpack

And got “Unable to locate package nvidia-jetpack no packages found”

5

Hi,

Please check this with cat /etc/nv_tegra_release.
For JetPack 5.0.2, the BSP version is r35.1.

Thanks.

6

It is R35. So that means that I should be using FROM “nvcr.io/nvidia/l4t-tensorrt:r8.4.1.5-devel”? That is the one I have been

8

I have the build working with “FROM nvcr.io/nvidia/l4t-tensorrt:r8.4.1.5-devel”. I am now trying to start it with “up” and I get this error (I believe this has the relevant parts). Any ideas?:

app | [08/18/2024-20:17:07] [TRT] [E] 6: The engine plan file is not compatible with this version of TensorRT, expecting library version 8.4.1.5 got 8.4.2.4, please rebuild.
app | [08/18/2024-20:17:07] [TRT] [E] 4: [runtime.cpp::deserializeCudaEngine::49] Error Code 4: Internal Error (Engine deserialization failed.)

9

Hi,

The error indicates the TensorRT engine is serialized with a different version.
Please convert the model with the current TensorRT (v8.4) again.

Thanks.

11

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