local host can't connect to notebooks on fresh install

[rayset]

this is what I run:

sudo GPU=0 ./agmb-docker run -d bethgelab/jupyter-deeplearning-x:cuda8.0-cudnn5

and this is the output

ret run -d bethgelab/jupyter-deeplearning-x:cuda8.0-cudnn5 run
1st CMD run

* Setting user name to:             root
* Setting user ID to:               0
* Setting user groups to:           bethgelab:1019,cin:1011,sudo
* Setting password to:              pw
* Setting SSH port binding to:      57293 (to set manually add -p 57293:22 as flag)
* Setting Notebook port binding to: 683 (to set manually add -p 683:8888 as flag)

You can now open the notebook on the host machine by directing your browser to

    http://localhost:683

or, from a remote system, to

    http://192.168.1.106:683

In the latter case make sure that your local machine can see the server! Otherwise you might have to configure an SSH tunnel first.

9737c301b1c51853e5101d2161032074a8de23353fc07060880f503287e68f01

I have to put the sudo before the command else way I get this error:

Error: Cannot connect to the Docker daemon. Is the docker daemon running on this host?

going to 192.168.1.106:683 gives me:

Firefox can't establish a connection to the server at 192.168.1.106:683.

    The site could be temporarily unavailable or too busy. Try again in a few moments.
    If you are unable to load any pages, check your computer's network connection.
    If your computer or network is protected by a firewall or proxy, make sure that Firefox is permitted to access the Web.

I'm on Ubuntu, gtx 1070 with cuda and cudnn installed.

what could be the problem?

thanks in advance!

[aecker]

Regarding the Docker daemon error: don't use sudo, make sure your user is in the docker group.

Regarding the problem with connecting to the notebook. Have you tried it from the host machine via localhost:683?

[rayset]

thanks for the reply!

yes, even localhost:683 and 127.0.0.1:683 return their equivalent pages.

I noticed that after plotting ''9737c301b1c51853e5101d2161032074a8de23353fc07060880f503287e68f01'' the terminal is ready to receive another command, while I was expecting some sort of ''busy state''. Don't know if this should be the case tho.

I'll look into the user groups now.

[aecker]

The -d flag means that the container is being detached. That's why you're getting the terminal back. Try removing the flag or inspecting the logs. Also, you can use ps aux | grep jupyter to check if the Jupyter server is running at all.

[wielandbrendel]

The Jupyter notebook is likely failing to start: agmb-docker mounts and sets the home directory to our internal /gpfs01/bethge/... address, fails to mount and so Jupyter fails to start in that directory. You will have to mount and set your home directory with -v $HOME:$HOME -e USER_HOME=$HOME yourself.

@aecker : That's a behaviour we could change - it should be enough to just mount and set the home directory of the user invoking the script, right?