Getting Started with Anvil

Once you’ve got your account configured it’s time to get started using Anvil!

The steps below provide a high level overview of some of the resources that we have available at Purdue.

OnDemand

The OnDemand tool is an open source platform for a multitude of things. At Purdue in The Data Mine we use it to launch our Jupyter Lab sessions.

Follow the instructions below to launch your first Jupyter Lab instance.

Be careful when changing resources on your instance. These high performance computing resources are shared by multiple users.

Making a very large request can have negative impacts on other users.

  1. Open your favorite browser, we usually use Firefox (not Chrome, not Edge, and not Safari), and navigate to ondemand.anvil.rcac.purdue.edu.

  2. Log-in with your ACCESS credentials that you set up previously.

  3. Authenticate with Duo.

  4. After you finish your authentication, there is a landing page for Anvil. This page has a dark-grey options bar across the top of the browser window with multiple options listed.

  5. Select the second to last option, which should say "The Data Mine". This will open a drop-down list for the different Data Mine appications on Anvil.

    The OnDemand Options Bar
    Figure 1. The OnDemand Options Bar

  6. Select the application that you would like to use from the drop-down list. For seminar projects, this would be the Jupyter Notebook application.

    • This will open a new screen that will give you options for how you want to start the application.

  7. An overview of the different application choices for the Jupyter Notebook application is included below:

    • Allocation - this is automatically populated with The Data Mine’s information. It only needs to be changed if you’d rather use GPU instead of CPU.

    • Queue - similar to Allocation, this only needs to be changed if you want to use GPU instead of CPU.

    • Time in Hours - This is the amount of time that the application will run for. It can be helpful to change this to a longer time.

      • For example, if you’re going to be working with a Jupyter Lab session in your Team’s lab meeting, you can change the value to 2 hours.

    • CPU Cores - This changes the number of cores and allocated memory that will be requested for your session.

      • If you notice that your code is frequently crashing, it may be beneficial to increase the number of cores.

  8. If you have all of your settings updated as needed, you can click the blue Launch button at the bottom of the screen.

  9. Once the session is ready you’ll see a link to connect to the Jupyter Notebook.

    Jupyter Notebook Launch
    Figure 2. Jupyter Notebook Launch

  10. In the window representing the running Jupyter Notebook task, select Connect to Jupyter to open the Jupyter Notebook in a new window.

Pay careful attention to your time remaining once the cluster has started.

Once the time runs out the session will end and you cannot save any new changes.

Helpful Tips

  • It is helpful to create a symlink to your team’s folder in Jupyter Lab.

    • To do this, open a cell in Jupyter Lab, and type this code exactly (be sure to check it carefully, including the spaces).

    • NOTE You only need to run this command one time (ever)! If you run it more than once Anvil will get angry.

      %%bash

ln -s /anvil/projects/tdm/corporate $HOME