3. Customize Attributes¶

Now we will customize the app to work on a given cluster. Be sure that you walk through Software Requirements for the given cluster ahead of time.

3.1. Modify Form File¶

The main responsibility of the form.yml file located in the root of the app is for defining the attributes (their values or HTML form elements) used when generating the batch script.

We will begin by adding a cluster for the Jupyter app to use. You do this by editing the form.yml in your favorite editor as such:
```
# form.yml
---

...

cluster: "my_cluster"

...
```
where we replace my_cluster with a valid cluster that corresponds to a cluster configuration file located under /etc/ood/config/clusters.d/my_cluster.yml.
We will also edit the module(s) that is(are) required to be loaded within our batch job to get a Jupyter Notebook Server running:
```
# form.yml
---

...

attributes:

  ...

  modules: "python/3.5"

...
```
where we replace python/3.5 with a list of required modules for our given cluster.
Optional If we do NOT have the Conda extensions installed for the above Python modules then we must disable it:
```
# form.yml
---

...

attributes:

  ...

  conda_extensions: "0"

...
```
They are enabled by default in this file.

Note

It is recommended you commit the changes you made to form.yml to git:

# Stage the modified file and commit your changes
git commit form.yml -m 'updated form with cluster attributes'

3.2. Verify it Works¶

You can now test the app again by visiting your local OnDemand server in your browser:

GET /pun/sys/dashboard/batch_connect/dev/jupyter_app/session_contexts/new HTTP/1.1
Host: ondemand.my_center.edu

You should see a web form for the Jupyter app. Fill in the form now and try to “Launch” a Jupyter batch job.

Note

While you are waiting for the job to start it is recommended that you click the link under the “Session ID:”. This will open the File Explorer in the working directory of the currently launched Jupyter batch job.

Useful debugging files (before job runs):

user_defined_context.json - Attributes submitted by the user in the web form.
job_script_content.sh - The batch script content.
job_script_options.json - The job submission parameters (this will be used in the next section if you have trouble submitting the job).

Useful debugging files (after job runs):

output.log - This is the log file of the batch job. This is helpful to look at if your batch job dies abruptly due to an invalid module or missing Jupyter libraries.

Continue to the next section to learn about job submission parameters.

Warning

The app will probably display a warning about requiring a cluster. This is perfectly fine. Continue on to the next section to learn about customizing the app.