install failure of TensorRT

[Tom-Teamo]

Description

When I install TrensorRT in my Ubuntu22.04 as step No.1 in setup, it always failed at "extracting: ngc-cli.md5 ". And I do not know how to fix this.

... 
9.464   inflating: ngc-cli/opentelemetry/proto/common/v1/common_pb2.pyi  
9.464   inflating: ngc-cli/opentelemetry/proto/version.py  
9.464    creating: ngc-cli/opentelemetry/util/
9.464   inflating: ngc-cli/opentelemetry/util/_once.py  
9.464    creating: ngc-cli/opentelemetry/util/__pycache__/
9.464   inflating: ngc-cli/opentelemetry/util/__pycache__/_once.cpython-39.pyc  
9.464   inflating: ngc-cli/opentelemetry/util/__pycache__/_providers.cpython-39.pyc  
9.464   inflating: ngc-cli/opentelemetry/util/__pycache__/re.cpython-39.pyc  
9.464   inflating: ngc-cli/opentelemetry/util/__pycache__/_importlib_metadata.cpython-39.pyc  
9.464   inflating: ngc-cli/opentelemetry/util/__pycache__/types.cpython-39.pyc  
9.464   inflating: ngc-cli/opentelemetry/util/_importlib_metadata.py  
9.464   inflating: ngc-cli/opentelemetry/util/_providers.py  
9.464   inflating: ngc-cli/opentelemetry/util/re.py  
9.464   inflating: ngc-cli/opentelemetry/util/types.py  
9.464  extracting: ngc-cli.md5             
71.87 Connection failed; retrying... (Retries left: 5)
137.0 Connection failed; retrying... (Retries left: 4)
202.1 Connection failed; retrying... (Retries left: 3)
267.3 Connection failed; retrying... (Retries left: 2)
332.4 Connection failed; retrying... (Retries left: 1)
332.4 Error: Request timed out.
------
ubuntu-20.04.Dockerfile:109
--------------------
 107 |     
 108 |     # Download NGC client
 109 | >>> RUN cd /usr/local/bin && wget https://ngc.nvidia.com/downloads/ngccli_cat_linux.zip && unzip ngccli_cat_linux.zip && chmod u+x ngc-cli/ngc && rm ngccli_cat_linux.zip ngc-cli.md5 && echo "no-apikey\nascii\n" | ngc-cli/ngc config set
 110 |     
 111 |     # Set environment and working directory
--------------------
ERROR: failed to solve: process "/bin/bash -c cd /usr/local/bin && wget https://ngc.nvidia.com/downloads/ngccli_cat_linux.zip && unzip ngccli_cat_linux.zip && chmod u+x ngc-cli/ngc && rm ngccli_cat_linux.zip ngc-cli.md5 && echo \"no-apikey\\nascii\\n\" | ngc-cli/ngc config set" did not complete successfully: exit code: 1

Environment

TensorRT Version: latest

NVIDIA GPU: 3090

NVIDIA Driver Version: 12.2

CUDA Version: 12.1

Operating System: Ubuntu22.04

Steps To Reproduce

just as step No.1 in setup, it always failed at "extracting: ngc-cli.md5 ".

[lix19937]

Maybe network problem, you need a stable vpn

[YongWookHa]

I'm facing same problem at extracting: ngc-cli.md5.
@lix19937 did you test it?

[lix19937]

you can offline download ngccli_cat_linux.zip and then unzip it and then
chmod u+x ngc-cli/ngc && echo \"no-apikey\\nascii\\n\" | ngc-cli/ngc config set

[YongWookHa]

@lix19937 It's not about network problem.
The command stucks at the echo "no-apikey\nascii\n" | ngc-cli/ngc config set part.

[YongWookHa]

I've just resolved the problem by the steps below.

1. Split the line in dockerfile

TensorRT/docker/ubuntu-20.04.Dockerfile

Line 113 in c0c633c

RUN cd /usr/local/bin && wget https://ngc.nvidia.com/downloads/ngccli_cat_linux.zip && unzip ngccli_cat_linux.zip && chmod u+x ngc-cli/ngc && rm ngccli_cat_linux.zip ngc-cli.md5 && echo "no-apikey\nascii\n" | ngc-cli/ngc config set

to

# Download NGC client
RUN cd /usr/local/bin && wget https://ngc.nvidia.com/downloads/ngccli_cat_linux.zip
RUN cd /usr/local/bin && unzip ngccli_cat_linux.zip
RUN cd /usr/local/bin && chmod u+x ngc-cli/ngc
RUN cd /usr/local/bin && rm -f ngccli_cat_linux.zip ngc-cli.md5
RUN cd /usr/local/bin && echo "no-apikey\nascii\n" | ngc-cli/ngc config set

2. Comment out the problem causing command

# RUN cd /usr/local/bin && echo "no-apikey\nascii\n" | ngc-cli/ngc config set

3. Build the docker image

> ./docker/build.sh --file docker/ubuntu-20.04.Dockerfile --tag tensorrt-ubuntu20.04-cuda12.1

4. Launch the container

> ./docker/launch.sh --tag tensorrt-ubuntu20.04-cuda12.1 --gpus all

5. Put the command manually

> ngc-cli/ngc config set
Enter API key [no-apikey]. Choices: [<VALID_APIKEY>, 'no-apikey']: no-apikey
Enter CLI output format type [ascii]. Choices: ['ascii', 'csv', 'json']: ascii
Validating configuration...
Successfully validated configuration.
Saving configuration...
Successfully saved NGC configuration to /home/trtuser/.ngc/config

[guiyuliu]

hello，do you know why this step was stucked ？ “echo "no-apikey\nascii\n" | ngc-cli/ngc config set”
I encountered the same problem here

[bashirmindee]

doing echo "no-apikey\Nascii\n" | ... is really weird.
I suggest doing: /usr/local/bin/ngc-cli/ngc config set --format_type ascii

[bashirmindee]

for some reason, you have to run ngc-cli/ngc config twice in order for it to work!
The 1st time it blocks and doesn't work. The second time it works correctly

[bashirmindee]

I finally found it.
ngc config ... generates /root/.ngc/meta_data if it doesn't exist. However it blocks the 1st time because it doesn't already exist.
by generating dummy data you can bypass this:

RUN mkdir -p /root/.ngc
RUN echo -e ";WARNING - This is a machine generated file.  Do not edit manually.\n\n[COMMAND_MAP]\n\n[UPGRADE]\nlast_upgrade_msg_date = 2024-03-26::09:43:31\n\n[USER_ROLES]\n\n[PRODUCT_NAMES]\n" > /root/.ngc/meta_data
RUN /usr/local/bin/ngc-cli/ngc config set --format_type ascii

This works directly. I can do a PR to fix it

[Goodluckhf]

It happened after release new CLI version. I installed specific previous version and it fixed.

you can change this line

RUN cd /usr/local/bin && wget https://ngc.nvidia.com/downloads/ngccli_cat_linux.zip && unzip ngccli_cat_linux.zip && chmod u+x ngc-cli/ngc && rm ngccli_cat_linux.zip ngc-cli.md5 && echo "no-apikey\nascii\n" | ngc-cli/ngc config set

To ->

RUN cd /usr/local/bin && wget --content-disposition https://api.ngc.nvidia.com/v2/resources/nvidia/ngc-apps/ngc_cli/versions/3.38.0/files/ngccli_linux.zip -O ngccli_linux.zip && unzip ngccli_linux.zip && chmod u+x ngc-cli/ngc && rm ngccli_linux.zip ngc-cli.md5 && echo "no-apikey\nascii\n" | ngc-cli/ngc config set

p.s you can choose specific version here -> https://org.ngc.nvidia.com/setup/installers/cli

[Nghiauet]

@Goodluckhf Thanks it work for me.