Nvidia-docker: Support latest docker-ce on Debian stretch

[davidwessman]

1. Issue or feature description

Upgraded to docker-ce 17.12.1~ce~debian and nvidia-docker2 is pinned to docker-ce 17.12.0~ce~debian in this Makefile.

I can therefore not install nvidia-docker2 without downgrading docker-ce as far as I can see.

2. Steps to reproduce the issue

• Upgrade to latest docker-ce on Debian stretch.
• Try to install nvidia-docker2

Edit

I realised you fixed this 4 hours ago... ;)
When will the release be available?

[flx42]

The new packages should be released today (PST), just waiting to test one more thing with @3XX0

[Fish2]

@flx42 @3XX0 Any news on the release wondering if i should downgrade my docker ?

[flx42]

@Fish2 sorry, pushed back to Thursday or Friday.

[Fish2]

@flx42 ok thanks for letting me know

[ghisvail]

Same here for the Ubuntu packages. Is there a particular reason why nvidia-docker has to be pinned at a specific Docker version. Version 17.12.1 does not feature that much changes compared to 17.12.0.

[alexwitherspoon]

Any updates on the release? Thanks!

[flx42]

Done @alexwitherspoon, please try

[alexwitherspoon]

Confirmed working on Debian thanks @flx42 !

root@core:~# nvidia-docker version
NVIDIA Docker: 2.0.3
Client:
 Version:       17.12.1-ce
 API version:   1.35
 Go version:    go1.9.4
 Git commit:    7390fc6
 Built: Tue Feb 27 22:17:21 2018
 OS/Arch:       linux/amd64

Server:
 Engine:
  Version:      17.12.1-ce
  API version:  1.35 (minimum version 1.12)
  Go version:   go1.9.4
  Git commit:   7390fc6
  Built:        Tue Feb 27 22:15:56 2018
  OS/Arch:      linux/amd64
  Experimental: false

root@core:~# lsb_release -a
No LSB modules are available.
Distributor ID: Debian
Description:    Debian GNU/Linux 9.3 (stretch)
Release:        9.3
Codename:       stretch
root@core:~# uname -a
Linux <hostname> 4.9.0-6-amd64 #1 SMP Debian 4.9.82-1+deb9u2 (2018-02-21) x86_64 GNU/Linux

root@core:~# docker run --runtime=nvidia --rm nvidia/cuda nvidia-smi
Unable to find image 'nvidia/cuda:latest' locally
latest: Pulling from nvidia/cuda
1be7f2b886e8: Already exists 
6fbc4a21b806: Already exists 
c71a6f8e1378: Already exists 
4be3072e5a37: Already exists 
06c6d2f59700: Already exists 
5902457f349a: Pull complete 
fbd0792faac4: Pull complete 
f001f17adb25: Pull complete 
4acb8434e54f: Pull complete 
8c157ee3954c: Pull complete 
Digest: sha256:288b4d2f9aa1f75c8c6e623a1d305bf2f66078797f11d03bd4720a4fdaaf534b
Status: Downloaded newer image for nvidia/cuda:latest
Sun Mar  4 13:38:22 2018       
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 384.81                 Driver Version: 384.81                    |
|-------------------------------+----------------------+----------------------+
| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |
| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |
|===============================+======================+======================|
|   0  Quadro P4000        Off  | 00000000:0E:00.0 Off |                  N/A |
| 59%   72C    P0   104W / 105W |   2541MiB /  8113MiB |    100%      Default |
+-------------------------------+----------------------+----------------------+

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

[Fish2]

This is back again with the new release of docker-ce

[flx42]

@Fish2 what are you seeing? I've released packages this morning

[Fish2]

@flx42 just checked and its all good now. ubuntu wanted to remove Nvidia as there was not update available to match docker-ce

[flx42]

@Fish2 ok! thanks for confirming