Introducing fastai-shell

I managed to get an instance running (P4) in us-west2-c. however, I cannot connect using the given IP. Do I need to change stuff in GCP settings?

Thanks. This is very useful.
I followed all the steps but I am not able to access http://my-external-ip:8888. Even after running a new instance of jupyter notebook on port 8889, I cannot access it until I change the NoteBookApp.IP in jupyter_notebook_config.py from ā€˜localhost’ to ā€˜0.0.0.0’.

In lsof output I can see even port 8888 is bound on localhost.

(fastai-v1) nutane@fastai:~$ lsof |grep 8888
jupyter-n 1625 nutane 6u IPv4 22390 0t0 TCP localhost:8888 (LISTEN)

Am I doing anything wrong? BTW, why doesn’t this instance of jupyter notebook list under jupyter notebook list command?

Did you try port: 8080 ?
In the latest script I changed it. Anyway, you’ll see the correct URL after it’s started in the CLI.

Port number was not the problem. Bound IP address was. I did not use the fastai shell. Rather I used the gcloud compute command as I was getting some error for fastai-shell.
For GCP instance, does it make sense to bind to ā€˜localhost’ in default jupyter config? I’d think 0.0.0.0 would be better option.

If you bind to localhost you can’t access it from outside. Use 0.0.0.0.

Yup, that’s exactly my point. The default jupyter config should also have 0.0.0.0. Anyways, I am up and running with fastai-shell. Earlier in the day the commands where throwing weird errors because of which I tried manual options (which gave me quite a bit of pain). Good job. Thanks.

Hi again.

I finally managed to spinup an instance again. But now I do get a 403 error when I try to connect…

http://35.230.109.121:8080

Is there a way to check what is going on? Do I need to do some network setup in my project settings in GCP?

Could that be a conda thing on the instance ?

Thanks,
C

That’s pretty interesting. I’ll have a look.

1 Like

thanks!

I had an issue with the official image.
Got the same issue.

So, I went back to custom image.
Could you try to update fastai-shell and try again.
See: https://github.com/arunoda/fastai-shell#stay-up-to-date

I just updated to 0.1 before tried it… Do you know if I should check project settings in GCP for network issues?

Tried it again: same result… 403 forbidden

I just updated to 0.1 before tried it

That’s interesting. Let’s me try it again.

One thing: You’ve to run fastai destroy and fastai create again.
You have to upload any crucial data before doing this.

Very sorry for this.
But I couldn’t rely on the official image since it’s getting changed and I have no control over that.
So starting scratch from the base ubuntu is method which will work for sure.

Ok, did that and finally have a connection to GCP! Man, I was getting desperate :wink:

1 Like

hello
i am looking to download some datasets straight to gcp, unzip them and use them in a fastai instance.
can you please suggest how to do this?
thank you very much

Once you start server. You can access your notebooks from https://ip:8080.
In that you create a new Terminal.
Then you can download them and unzip.

Hi @arunoda,

I am now using fastai-shell and it is working beautifully. Thanks again.
Some questions, does the image used here differ considerably from the official one?
I wanted to try using Jeremy’s .vimrc but ran into quite a bit of issues. The vim did not have the python support which I then compiled in but still getting other issues.

It’s a different setup. But it should be the same.
Try following before you run any commands:

source activate fastai-v1

I’ve been using fastai-shell and think it works great! Thanks a lot for putting it together @arunoda

There’s only one are where I struggle that is the nbextensions. I have used:

`conda install -c conda-forge jupyter_contrib_nbextensions`
`conda install -c conda-forge jupyter_nbextensions_configurator`

in the jupyter terminal, and have restarted the instance, but for some reason I cannot still see the jupyter_nbextensions_configurator tab.

I have checked the jupyter serverextension list:

oguiza@fastai-1:~$ jupyter serverextension list
config dir: /home/oguiza/.jupyter
    jupyter_nbextensions_configurator  enabled
    - Validating...
      jupyter_nbextensions_configurator 0.4.0 OK
    jupyter_notebook_gist  enabled
    - Validating...
      jupyter_notebook_gist 0.5.0 OK
config dir: /home/oguiza/anaconda3/etc/jupyter
    jupyterlab  enabled
    - Validating...
      jupyterlab 0.34.9 OK
    jupyter_nbextensions_configurator  enabled
    - Validating...
      jupyter_nbextensions_configurator 0.4.0 OK

and jupyter nbextension list:

    oguiza@fastai-1:~$ jupyter nbextension list
Known nbextensions:
  config dir: /home/oguiza/.jupyter/nbconfig
    common section
      jupyter-notebook-gist/common  enabled
      - Validating: problems found:
        - require?  X jupyter-notebook-gist/common
    notebook section
      nbextensions_configurator/config_menu/main  enabled
      - Validating: problems found:
        - require?  X nbextensions_configurator/config_menu/main
      contrib_nbextensions_help_item/main  enabled
      - Validating: OK
      jupyter-js-widgets/extension  enabled
      - Validating: OK
      jupyter-notebook-gist/notebook-extension  enabled
      - Validating: problems found:
        - require?  X jupyter-notebook-gist/notebook-extension
    tree section
      nbextensions_configurator/tree_tab/main  enabled
      - Validating: problems found:
        - require?  X nbextensions_configurator/tree_tab/main
      jupyter-notebook-gist/tree-extension  enabled
      - Validating: problems found:
        - require?  X jupyter-notebook-gist/tree-extension
  config dir: /home/oguiza/anaconda3/etc/jupyter/nbconfig
    notebook section
      jupyter-js-widgets/extension  enabled
      - Validating: OK
      nbextensions_configurator/config_menu/main  enabled
      - Validating: problems found:
        - require?  X nbextensions_configurator/config_menu/main
      contrib_nbextensions_help_item/main  enabled
      - Validating: OK
    tree section
      nbextensions_configurator/tree_tab/main  enabled
      - Validating: problems found:
        - require?  X nbextensions_configurator/tree_tab/main

I have also uninstalled and reinstalled it again, but unsuccessfully. Could anyone please help?

When you run your terminal, activate the conda env like this:

source activate fastai-v1

I think I’m going to add this automatically.

1 Like