SDK Manager Install DRIVE Failure

[Peter_Pertrili]

Our environment is Ubuntu 18.04 and sdkmanager_1.5.0-7774_amd64.deb. When using SDK Manager to install DRIVE, we meet relative error message below

08:30:35 INFO: CUDA on Host - host: Fetched 177 kB in 4s (42.7 kB/s)
08:30:36 INFO: CUDA on Host - host: Reading package lists...
08:30:36 INFO: CUDA on Host - host: Reading package lists...
08:30:37 INFO: CUDA on Host - host: Building dependency tree...
08:30:37 INFO: CUDA on Host - host: Reading state information...
08:30:37 INFO: CUDA on Host - host: The following additional packages will be installed:
08:30:37 INFO: CUDA on Host - host: cuda-command-line-tools-10-2 cuda-compiler-10-2 cuda-cudart-10-2
08:30:37 INFO: CUDA on Host - host: cuda-cudart-dev-10-2 cuda-cufft-10-2 cuda-cufft-dev-10-2 cuda-cuobjdump-10-2
08:30:37 INFO: CUDA on Host - host: cuda-cupti-10-2 cuda-cupti-dev-10-2 cuda-curand-10-2 cuda-curand-dev-10-2
08:30:37 INFO: CUDA on Host - host: cuda-cusolver-10-2 cuda-cusolver-dev-10-2 cuda-cusparse-10-2
08:30:37 INFO: CUDA on Host - host: cuda-cusparse-dev-10-2 cuda-documentation-10-2 cuda-driver-dev-10-2
08:30:37 INFO: CUDA on Host - host: cuda-gdb-10-2 cuda-libraries-10-2 cuda-libraries-dev-10-2 cuda-license-10-2
08:30:37 INFO: CUDA on Host - host: cuda-memcheck-10-2 cuda-misc-headers-10-2 cuda-npp-10-2 cuda-npp-dev-10-2
08:30:37 INFO: CUDA on Host - host: cuda-nsight-10-2 cuda-nsight-compute-10-2 cuda-nsight-systems-10-2
08:30:37 INFO: CUDA on Host - host: cuda-nvcc-10-2 cuda-nvdisasm-10-2 cuda-nvgraph-10-2 cuda-nvgraph-dev-10-2
08:30:37 INFO: CUDA on Host - host: cuda-nvjpeg-10-2 cuda-nvjpeg-dev-10-2 cuda-nvml-dev-10-2 cuda-nvprof-10-2
08:30:37 INFO: CUDA on Host - host: cuda-nvprune-10-2 cuda-nvrtc-10-2 cuda-nvrtc-dev-10-2 cuda-nvtx-10-2
08:30:37 INFO: CUDA on Host - host: cuda-nvvp-10-2 cuda-samples-10-2 cuda-sanitizer-api-10-2 cuda-tools-10-2
08:30:37 INFO: CUDA on Host - host: cuda-visual-tools-10-2
08:30:37 INFO: CUDA on Host - host: The following held packages will be changed:
08:30:37 INFO: CUDA on Host - host: cuda-toolkit-10-2
08:30:37 INFO: CUDA on Host - host: The following packages will be upgraded:
08:30:37 INFO: CUDA on Host - host: cuda-command-line-tools-10-2 cuda-compiler-10-2 cuda-cudart-10-2
08:30:37 INFO: CUDA on Host - host: cuda-cudart-dev-10-2 cuda-cufft-10-2 cuda-cufft-dev-10-2 cuda-cuobjdump-10-2
08:30:37 INFO: CUDA on Host - host: cuda-cupti-10-2 cuda-cupti-dev-10-2 cuda-curand-10-2 cuda-curand-dev-10-2
08:30:37 INFO: CUDA on Host - host: cuda-cusolver-10-2 cuda-cusolver-dev-10-2 cuda-cusparse-10-2
08:30:37 INFO: CUDA on Host - host: cuda-cusparse-dev-10-2 cuda-documentation-10-2 cuda-driver-dev-10-2
08:30:37 INFO: CUDA on Host - host: cuda-gdb-10-2 cuda-libraries-10-2 cuda-libraries-dev-10-2 cuda-license-10-2
08:30:37 INFO: CUDA on Host - host: cuda-memcheck-10-2 cuda-misc-headers-10-2 cuda-npp-10-2 cuda-npp-dev-10-2
08:30:37 INFO: CUDA on Host - host: cuda-nsight-10-2 cuda-nsight-compute-10-2 cuda-nsight-systems-10-2
08:30:37 INFO: CUDA on Host - host: cuda-nvcc-10-2 cuda-nvdisasm-10-2 cuda-nvgraph-10-2 cuda-nvgraph-dev-10-2
08:30:37 INFO: CUDA on Host - host: cuda-nvjpeg-10-2 cuda-nvjpeg-dev-10-2 cuda-nvml-dev-10-2 cuda-nvprof-10-2
08:30:37 INFO: CUDA on Host - host: cuda-nvprune-10-2 cuda-nvrtc-10-2 cuda-nvrtc-dev-10-2 cuda-nvtx-10-2
08:30:37 INFO: CUDA on Host - host: cuda-nvvp-10-2 cuda-samples-10-2 cuda-sanitizer-api-10-2 cuda-toolkit-10-2
08:30:37 INFO: CUDA on Host - host: cuda-tools-10-2 cuda-visual-tools-10-2
08:30:37 INFO: CUDA on Host - host: 46 upgraded, 0 newly installed, 0 to remove and 58 not upgraded.
08:30:37 ERROR: CUDA on Host - host: E: Held packages were changed and -y was used without --allow-change-held-packages.
08:30:37 INFO: CUDA on Host - host: [ Package Install Finished with Error ]
08:30:37 INFO: CUDA on Host - host: [host] [ 2.55 GB used. Disk Avail: 369.02 GB ]
08:30:37 INFO: CUDA on Host - host: [ NV_CUDA_HOST_COMP Install took 36s ]
08:30:37 ERROR: CUDA on Host - host: [error]: Error when apt install failed; [exec_command]: sudo apt-get update; sleep 0.5; sudo apt-get -y --allow-downgrades install cuda-toolkit-10-2=10.2.187* ; [error]: exit status 100; [deb_path]: /home/xxxxx/Downloads/nvidia/sdkm_downloads/cuda-repo-ubuntu1804-10-2-local-10.2.187-440.101_1.0-1_amd64.deb

08:30:37 ERROR: CUDA on Host - host: command terminated with error
08:30:37 SUMMARY: CUDA on Host - host: First Error: Failed to install debian repository file [host]

OSS Packets Error Message

9:17:28 INFO: OSS Packages - target_image: rsync was already set to manually installed.
09:17:28 INFO: OSS Packages - target_image: Package mount is already installed, Setting it to manually installed
09:17:29 INFO: OSS Packages - target_image: mount was already set to manually installed.
09:17:29 INFO: OSS Packages - target_image: Package tar is already installed, Setting it to manually installed
09:17:29 INFO: OSS Packages - target_image: tar was already set to manually installed.
09:17:29 INFO: OSS Packages - target_image: Non interactive : Installing package linux-libc-dev:i386
09:17:29 INFO: OSS Packages - target_image: Reading package lists...
09:17:29 INFO: OSS Packages - target_image: Building dependency tree...
09:17:29 INFO: OSS Packages - target_image: Reading state information...
09:17:30 ERROR: OSS Packages - target_image: E: Unable to locate package linux-libc-dev:i386
09:17:30 INFO: OSS Packages - target_image: ERROR: linux-libc-dev:i386 failed to install
09:17:30 INFO: OSS Packages - target_image: Non interactive mode on : Triggering exit
09:17:30 INFO: OSS Packages - target_image: Exiting.
09:17:30 ERROR: OSS Packages - target_image: ERROR: Installation failed.
09:17:30 INFO: OSS Packages - target_image: [ Package Install Finished with Error ]
09:17:30 INFO: OSS Packages - target_image: [host] [ 875.92 MB used. Disk Avail: 357.33 GB ]
09:17:30 INFO: OSS Packages - target_image: [ NV_OSS_PACKAGES_SDK_COMP Install took 17s ]
09:17:30 ERROR: OSS Packages - target_image: [error]: [exit_code]: 1: [exec_command]: /bin/bash -c /home/xxxxx/Downloads/nvidia/sdkm_downloads/drive-t186ref-linux-5.2.0.0-23067027-oss-minimal-sdk.run --nox11 | tee -a /home/xinping/.nvsdkm/logs/DRIVE_OS_5.2.0_SDK_Linux_OS/NV_OSS_PACKAGES_SDK_COMP.log; exit ${PIPESTATUS[${#PIPESTATUS[@]}-2]}; [error_message]: ERROR: Installation failed.
09:17:30 ERROR: OSS Packages - target_image: command terminated with error
09:17:30 SUMMARY: OSS Packages - target_image: Unable to locate debian package

I also found similar thread below:

SDK Manager，can not install cuda and cudnn

Hi, When I use SDK Manager to install cuda and cudnn ,I find it always fail. below is the error: 22:33:29 INFO : CUDA Toolkit for L4T : Unhandled error when running sudo apt-get -y update 22:33:29 INFO : CUDA Toolkit for L4T : exit status 1 22:33:29 INFO : CUDA Toolkit for L4T : [ Package Install Finished with Error ] 22:33:29 ERROR : CUDA Toolkit for L4T : NV_L4T_CUDA_TARGET_POST_INSTALL_COMP command scp -q -o UserKnownHostsFile=/dev/null -o StrictHostKeyChecking=no /tmp/tmp_NV_L4T_CUDA_TARG…

Error when install "cuda host( and other package)" by SDK manager

I’m trying to follow the installation instructions on this link to install the Host Components on my Ubuntu computer (and also the Jetson Nano after this) , and I get the following error message. Any thoughts, please? Thank you in advance! Link: https://docs.nvidia.com/sdk-manager/install-with-sdkm-jetson/index.html My host computer setup & error screen shot: Ubuntu16.04 driver: GTX1660 (driver version 440) no cuda pre-installed [Screenshot] My Terminal & error: 17:25:57 INFO : CUDA on…

But it doesn’t work for me.
Can you give me some advice?

[Peter_Pertrili]

Even CUDA Cross-Compile For Linux still installed failure.

09:32:56 INFO: CUDA Cross-Compile for Linux - host: verifying checksum of /home/xxxx/Downloads/nvidia/sdkm_downloads/cuda-repo-cross-aarch64-10-2-local-10.2.187_1.0-1_all.deb
09:32:58 INFO: CUDA Cross-Compile for Linux - host: Found file /home/xxxx/Downloads/nvidia/sdkm_downloads/cuda-repo-cross-aarch64-10-2-local-10.2.187_1.0-1_all.deb with correct checksum, skip downloading.
09:33:23 SUMMARY: CUDA Cross-Compile for Linux - host: Depends on failed component
09:57:11 INFO: CUDA Cross-Compile for Linux - host: verifying checksum of /home/xxxx/Downloads/nvidia/sdkm_downloads/cuda-repo-cross-aarch64-10-2-local-10.2.187_1.0-1_all.deb
09:57:11 INFO: CUDA Cross-Compile for Linux - host: Found file /home/xxxx/Downloads/nvidia/sdkm_downloads/cuda-repo-cross-aarch64-10-2-local-10.2.187_1.0-1_all.deb with correct checksum, skip downloading. 

What is meaning host: Depends on failed component?

But if I use dpkg to do force install cuda-repo deb packet, it works.

~/Downloads/nvidia/sdkm_downloads$ sudo dpkg -i cuda-repo-cross-aarch64-10-2-local-10.2.187_1.0-1_all.deb 
Selecting previously unselected package cuda-repo-cross-aarch64-10-2-local-10.2.187.
(Reading database ... 330546 files and directories currently installed.)
Preparing to unpack cuda-repo-cross-aarch64-10-2-local-10.2.187_1.0-1_all.deb ...
Unpacking cuda-repo-cross-aarch64-10-2-local-10.2.187 (1.0-1) ...
Setting up cuda-repo-cross-aarch64-10-2-local-10.2.187 (1.0-1) ...

[EdwardZhou]

Hi Peter,

The log is incomplete so it is hard to tell what’s the root cause.
One possible reason is that there are cuda packages conflict. i.e., there are other version of CUDA packages that conflict with the one to be installed in DRIVE OS 5.2.0.
Please uninstall all cuda packages and make sure the result of dpkg -l | grep cuda is empty then retry.
If this still doesn’t work, can you please provide the full log by packaging and uploading the ~/.nvsdkm folder for more investigation?

[Peter_Pertrili]

Hi EdwardZhou,

Seems my nvidia-smi already install CUDA 11.2, I need to downgrade CUDA11.2 and try again.

:~/.nvsdkm$ nvidia-smi 
Tue May 11 10:29:14 2021       
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 460.32.03    Driver Version: 460.32.03    CUDA Version: 11.2     |
|-------------------------------+----------------------+----------------------+
| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |
| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |
|                               |                      |               MIG M. |
|===============================+======================+======================|
|   0  GeForce RTX 2060    On   | 00000000:01:00.0  On |                  N/A |
| N/A   46C    P8     6W /  N/A |    613MiB /  5934MiB |     17%      Default |
|                               |                      |                  N/A |
+-------------------------------+----------------------+----------------------+
                                                                               
+-----------------------------------------------------------------------------+
| Processes:                                                                  |
|  GPU   GI   CI        PID   Type   Process name                  GPU Memory |
|        ID   ID                                                   Usage      |
|=============================================================================|
|    0   N/A  N/A      1539      G   /usr/lib/xorg/Xorg                 24MiB |
|    0   N/A  N/A      1793      G   /usr/bin/gnome-shell               81MiB |
|    0   N/A  N/A      2912      G   /usr/lib/xorg/Xorg                305MiB |
|    0   N/A  N/A      3094      G   /usr/bin/gnome-shell               40MiB |
|    0   N/A  N/A      3596      G   ...AAAAAAAAA= --shared-files       35MiB |
|    0   N/A  N/A      3780      G   /usr/lib/firefox/firefox            1MiB |
|    0   N/A  N/A      3927      G   /usr/lib/firefox/firefox            1MiB |
|    0   N/A  N/A      3980      G   /usr/lib/firefox/firefox            1MiB |
|    0   N/A  N/A      4419      G   /usr/lib/firefox/firefox            1MiB |
|    0   N/A  N/A      4444      G   python3                            26MiB |
|    0   N/A  N/A      6158      G   /usr/lib/firefox/firefox           11MiB |
|    0   N/A  N/A      6238      G   ...gAAAAAAAAA --shared-files       54MiB |
|    0   N/A  N/A      6601      G   /usr/lib/firefox/firefox            1MiB |
|    0   N/A  N/A      7551      G   /usr/lib/firefox/firefox            1MiB |
|    0   N/A  N/A     11104      G   ...areRendererForSitePerProc       10MiB |

[Peter_Pertrili]

Hi EdwardZhou,

Maybe nvidia-smi CUDA version is not related with my current issue.
Anyway, I have attached nvsdkm.tar for you investigation.
nvsdkm.tar (1.9 MB)

Thanks for your help

[Peter_Pertrili]

After uninstall CUDA version and resolve source.list issue, then do retry again.
SDK Manager install successfully.

Thanks.