RIS Open On Demand

Compute Resources

Have questions or need help with compute, including activation or issues? Follow this link.
User Agreement

In order to be able to access Open On Demand, you have to ssh connect to the Compute Platform at least once.
You can find out how to do so via our Compute Quick Start Guide

Open On Demand Video

“An intuitive, innovative, and interactive interface to remote computing resources” -Open OnDemand Website
Open on Demand or OOD is a web browser based interface that connects to compute resources, in this case, the RIS Compute Platform.

Users can use a web browser to access the RIS Compute Platform.
Users can use any number of supported applications or incorporate their own.
Users can navigate the RIS Compute Platform and their computing needs in a GUI instead of via the command line.

The first thing users will need to do is to make sure they’ve logged onto the RIS Compute Platform at least once to initialize their account.
Users can find out how to do that in our quick start guide.
Once this is done, all users need to do is point their browser at https://compute1-ood-1.ris.wustl.edu/ .
Users will be sent to a default page upon connection.

In the top menu, there are multiple options of how to interact with the RIS Compute Platform. Each option has a descriptive section below.

This page is an interface for displaying the files and directories on the compute system.
This page defaults to the user’s home directory.
Due to OOD implementation restrictions, this cannot access storage1 or scratch1.

From this menu, you can see active jobs or use the job composer to launch new jobs.

This page simply lists the jobs that a user has running on the RIS Compute Platform.
This lists all jobs running for the user, not just those started through OOD.
This page is similar to the bjobs command on the command line.

The arrow on the left hand side of the job listing allows for viewing of the job details.

The information here is once again equivalent to what can be obtained via the bjobs command.
You can also delete or kill the job from this page, which would be the same as using the bkill command on the command line.

This page allows users to create and launch jobs that are not part of the included applications list.
This page operates like launching a bsub job using a .bsub job file.
Users can learn more about .bsub job files on our job examples page.

Under the new job option, users can create a job from multiple sources. - A default template - A specified path - An existing running or completed job.

Once a user has selected a source for their template for creating a job, an editor will become available.

In the editor users can determine the job file name and the job file location.
The editor also allows users to edit the job file right in the web browers by clicking the Open Editor button.
It is within this section that knowledge of how to create the .bsub job file mentioned previously will be necessary.

#!/bin/bash
#BSUB -q general
#BSUB -g /username/groupname
#BSUB -a 'docker(ubuntu)'

echo "Hello World"

If you are a member of more than one compute group, you will be prompted to specify an LSF User Group with the following line #BSUB -G group_name.
The -g option must be included here.
All options that are part of the bsub command are available to be used here. Each option must have its own #BSUB line.

To save the script after making what edits are necessary, click the save button at the top of the script editor page.
In the editor, users can also open up a terminal in the browser to work in the command line.
There is also an option to open the directory that is the location of the job script.
Once the job is ready, it can be submitted via the submit button.
If the job submits successfully, you will see a success message. Otherwise you will see an error message with information as to the problem.

This option allows for users to connect directly to different clusters via a terminal.
At this current time, only 1 cluster is available, Compute1.
Note that this still has access to all the Compute1 resources.
This is simply designed this way in case there are multiple resources available on a system.

From the Interactive Apps menu, you can either select an app directly or click on the page to go to the interactive apps.

Launching the interactive app of the user’s choice will bring up a job creation editor.

From here, you can select the resources you require for the job, as well as mounting storage1/scratch1 directories.

Not all fields are required but most are. If a field is required and left blank the job will give an error when launched telling users the field needs a value.

From this page users can see what interactive sessions they have currently running.

When there are sessions, they are listed here along with their status.
The first status is Queued and this means that the Compute Platform is waiting to launch the job on the system.

The next status is Starting and this status indicates that the job is in the process of launching.

The third status is Running and this status means that the job has launched and is ready to be connected to.

Clicking the connect button once the job is running will connect the user to the chosen application.
In this example Jupyter Notebook was chosen.

As can be seen, the app runs just as expected.
Once a user is done using a particular app, the job can be deleted or killed.
On the My Interactive Sessions page, a user can click the delete button to do this.
A confirmation box will pop up asking if the user is sure about killing the job.

Besides the applications listed earlier, users can start up a default job that launches into a desktop.
This is done by selecting the Compute RIS Desktop app from the Interactive Apps menu.
This desktop is configured using noVNC and does not have any special additions.

There are multiple locations from which a user can launch a terminal.
Exactly as expected, this is a terminal that is launched through the web-browser allowing users to do command line work.