manifest for nvidia/cuda:11.0-base-ubuntu20.04 not found: manifest unknown: manifest unknown

[johnny-rodriguez]

johnny-ubuntu:~$ lsb_release -a

Distributor ID: Ubuntu
Description: Ubuntu 22.04.2 LTS
Release: 22.04
Codename: jammy

No LSB modules are available.
Distributor ID: Ubuntu
Description: Ubuntu 22.04.2 LTS
Release: 22.04
Codename: jammy

When I run the following, I get an error:
johnny-ubuntu:~$ docker run --rm --gpus all nvidia/cuda:11.0-base-ubuntu20.04 nvidia-smi

Unable to find image 'nvidia/cuda:11.0-base-ubuntu20.04' locally
docker: Error response from daemon: manifest for nvidia/cuda:11.0-base-ubuntu20.04 not found: manifest unknown: manifest unknown.
See 'docker run --help'.

However, when I run the following, it looks perfectly fine:
johnny-ubuntu:~$ sudo docker run --rm --runtime=nvidia --gpus all nvidia/cuda:11.6.2-base-ubuntu20.04 nvidia-smi

Sun Mar 5 20:06:31 2023
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 515.86.01 Driver Version: 515.86.01 CUDA Version: 11.7 |
|-------------------------------+----------------------+----------------------+
| 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 NVIDIA GeForce ... Off | 00000000:01:00.0 Off | N/A |
| 0% 33C P8 14W / 350W | 5MiB / 12288MiB | 0% Default |
| | | N/A |
+-------------------------------+----------------------+----------------------+

+-----------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=============================================================================|
+-----------------------------------------------------------------------------+

I have an GeForce 3080 Ti. I have little experience with docker and nvidia drivers, so any help would be greatly appreciated.

UPDATE: Did the nvidia/cuda:11.0-base image get removed?

[elezar]

Yes, the image was removed. Please use the nvidia/cuda:11.0.3-base-${DIST} images instead. DIST can be (among others): ubuntu20.04, ubuntu18.04, centos7 (see the list of tags).

Note that this does not have to match the system where you are running the containers.

[kbpark102]

When i use nvidia/cuda:11.0-cudnn8-devel-ubuntu18.04, i used libcudnn8=8.0.4.30-1+cuda11.0
but i can't use libcudnn8=8.0.4.30-1+cuda11.0 in nvidia/cuda:11.0.3-cudnn8-devel-ubuntu18.04
please let me know the cudnn version for nvidia/cuda:11.0.3-cudnn8-devel-ubuntu18.04

[Undefinedefity]

You can try different registry mirror, like mine ( I don't know if this can help because this registry mirror is China but it works for me):

1. go to /etc/docker/daemon.json
2. change the content to be

{
  "registry-mirrors": [
    "https://hub-mirror.c.163.com",
    "https://mirror.baidubce.com"
  ]
}

3. run these two commands to restart docker

sudo systemctl daemon-reload
sudo systemctl restart docker

Note: if these new mirrors still don't have the image, mine daemon.json registry has one aliyun address but it's private, you can go to aliyun.com to look for one (it's free).

---

update: You can find the corresponding dockerfile by expolering thier gitlab commit history at nvidia/cuda gitlab: https://gitlab.com/nvidia/container-images/cuda.git

[dvdknaap]

incase people like me also had this problem.

This link will be more future proof

[elezar]

Please see:

• https://gitlab.com/nvidia/container-images/cuda/blob/master/doc/supported-tags.md
• https://gitlab.com/nvidia/container-images/cuda/-/blob/master/doc/support-policy.md

The lifetime of images is out of scope for this project, so please consider opening an issue there instead.