Can't install jupyter support for Anaconda

Options
mahmoud_shihab
mahmoud_shihab Partner, Dataiku DSS Core Designer, Dataiku DSS ML Practitioner, Dataiku DSS Adv Designer, Registered Posts: 42 Partner

Hi All,

I am currently running a RedHat 8 VM on a Windows machine. It is running a DSS 11.0.2 instance.

When I try to create a conda environment, I seem to be having issues with installing jupyter.

I tried using the reinstall jupyter support button, but that doesn't work

I tried to manually install ipykernel and ipython, but that doesn't work either.

I'm a little lost on what to do as I'm sure the anaconda is installed correctly on the VM.

Creating any python environment without conda works, but I would like to have the anaconda env creation working as well.


Operating system used: Red Hat 8 VM

Answers

  • mahmoud_shihab
    mahmoud_shihab Partner, Dataiku DSS Core Designer, Dataiku DSS ML Practitioner, Dataiku DSS Adv Designer, Registered Posts: 42 Partner
    Options

    It also seems creating a complete R environment with conda doesn't work either

    But unlike python, I can click reinstall jupyter support, and it will install it just fine

  • Turribeach
    Turribeach Dataiku DSS Core Designer, Neuron, Dataiku DSS Adv Designer, Registered, Neuron 2023 Posts: 1,716 Neuron
    Options

    Out of interest why do you want to have conda working under Dataiku? Both conda and pip do effectively the same thing (aka package managers) with the main difference being that conda can handle OS level dependancies and pip can't. Conda tends to be used on developer machine installs (ie when using the Anaconda distribution) since it's simpler to just let conda handle both Python and OS level packages on developer machines. However on servers I would argue that you should stay away from conda. We have yet to find a package we could not install via pip in Dataiku. As a Dataiku Administrator I much rather handle any OS level package dependencies from Python packages myself than letting conda do it behind my back as part of a code environment install. Installing an OS level package dependency could have a wide impact on other Python code environments or even the whole system itself. Therefore we include all required OS level package dependencies as part of your Dataiku build and we make sure we test any new required packages before we allow them into the Dataiku build. By only using pip to manage Python code environments in Dataiku we reduce the risks of unexpecting other code environments or the server itself.

  • mahmoud_shihab
    mahmoud_shihab Partner, Dataiku DSS Core Designer, Dataiku DSS ML Practitioner, Dataiku DSS Adv Designer, Registered Posts: 42 Partner
    Options

    Hi Turribeach,

    It's mainly to understand what I did wrong during my installation and deployment.

    It's for me to learn and grow.

  • Turribeach
    Turribeach Dataiku DSS Core Designer, Neuron, Dataiku DSS Adv Designer, Registered, Neuron 2023 Posts: 1,716 Neuron
    Options

    Can you please post proper error messages and exactly what commands are you running?

  • mahmoud_shihab
    mahmoud_shihab Partner, Dataiku DSS Core Designer, Dataiku DSS ML Practitioner, Dataiku DSS Adv Designer, Registered Posts: 42 Partner
    Options

    Hi Turribeach,

    I'm not really running any commands right now.

    I go in the code env panel under administration and just create the python or R env with the "use conda" box checked.

    Then when that fails, I try "reinstall jupyter support" but it doesn't work

    So then I try to install ipykernel and ipython using pip, but then it fails.

    The images are all in the first post.

Setup Info
    Tags
      Help me…