Narwhal User Guide 

Accessing Narwhal

Below are the steps for accessing Narwhal from your computer.

Please remember that this requires both an active CHPC Protected Environment account and a connection to the UofU network (if not on campus, please use the campus VPN).

 

1. Find the correct remote desktop app for your computer.
   
   1. If you are working from a Windows computer, go to 'Start' > 'All Programs' > 'Accessories' > 'Remote Desktop Connection'.
   2. If you are working from a Linux computer, use xfreerdp or the RDP client of your choice. 
   3. If you have a Mac, go to the Appstore and download 'Windows App' - this is the new Windows RDP app. 
2. Launch the RDP app you chose from step 1 and start a new remote connection to narwhal.chpc.utah.edu
3. In the username field, put your uNID as ad\uNID
4. Authenticate with your password. You will be prompted for DUO two-factor authorization - you can type 'push' to have the authorization sent to your phone. Once that is complete, you now have a successful remote connection to Narwhal. 

 

Accessing Your Project Data on Narhwal

When you connect to Narwhal and click on 'This PC', you will notice that there are many different drives available for you to use. This section will be an explanation of what those different drives are and which drive is best to use (and when!):

Please note that these drive mounts may change in the future. At the time this was written, this is how the drives were mounted. If you notice that some information on this page is out of date, please hit the red 'Provide feedback' button at the bottom of this page and let the CHPC know.

 

Narwhal Drive Mounts

Devices and Drives

• Local Disk (C:): This has the file system mounted from the Linux cluster, enabling you to follow the path to your project directory that the CHPC provides upon creation, i.e. \uufs\chpc.utah.edu\common\HIPAA\IRB_00123456. All folders under the C: drive are mounted for your convenience and are symbolic links to the folders under the R:, S:, T:, U:, and Z: drives.
• Scratch (E:): This is a scratch directory that can be used for storing intermediate files created for the duration of a job. This is local to Narwhal, meaning that it cannot be accessed from Redwood, the Linux cluster for the PE. The scratch drive is spinning disk, so it will have similar I/O performance to your home directory.
• Profile Directory (D:): A profile directory is similar to a home directory, but it is only local to Narwhal and is not accessible via Redwood, the Linux cluster for the PE. This is an SSD and will have better I/O performance than the home and project directories.
  • We do recommend that, when preparing to run any computational jobs on Narhwal (particularly with statistical applications), that you move your input files to your profile directory (D:). This is because many computational jobs can be sensitive to I/O and you will find much better I/O performance on the local SSD that is the profile directory (D: drive). After the job completes, you should then move output data back to your project folder and delete any intermediate files to save space for others.

Network Locations

• Project Directories (R:, S:, T:, U:): These are mounted from the Linux server and will include data from your projects. All work done in the PE should be saved to your project folders and all other locations should be considered temporary. These project folders typically begin with 'IRB_' or 'proj_'. Don't have access to a project folder that you need access to? Request access here. 
• Home Directory (Z:): This is mounted from the Linux server and is the same as your home directory on Redwood. This means that you can access the same files in your home directory on both Narwhal and Redwood.

Don't have access to the project folder (e.g. an IRB folder) that you need for your work? Please request access by going to this link.

 

Things to Consider (when using Narwhal)

File Systems and Storage Limitations

Narwhal is a separate server that is not connected to the Linux compute nodes. This means that Narwhal is limited to the specs of the server it resides on. Narwhal is governed by 'Fair Share' and, as such, we ask that all users limit their usage of the resources (memory and CPU) to half of what is currently available. This means that if only 128Gb of memory is available, you only use 64Gb at most for your job to allow room for other users on Narwhal. 

If you run into resource constraints by following the policy above, please contact the CHPC at helpdesk@chpc.utah.edu.

This means that it is best to clean up files you don't need in the C:, D:, and E: drives, as they are local to Narwhal and can utilize all of the storage available in Narwhal, causing Narwhal to slow down or freeze. If you still need the files in the C:, D:, and E: drives, then please move them to the R:, S:, T:, U:, and Z: drives, which are not local to Narwhal and have much more storage available.

IMPORTANT: To help address the lack of storage space on Narwhal, files on the C:, D:, and E: drives older than 60 days will be deleted. Please be sure to save your work to the mounted Home (Z:) or Projects (R:, S:, T:, and U:) locations so that you don’t fall victim to the automated purge script. In addition to this change, files will no longer go to the recycling bin when deleted but instead will be deleted permanently. 

Software Limitations

Due to the limitations in storage resouces available locally on Narwhal, we are making efforts to keep users from installing applications (such as Anaconda Python) on their profile or other locations, as these installations end up being duplicated across multiple user profiles and take up an unecessarily large amount of space. Many of these commonly installed programs, have been installed within Program Files so that all users may access it. 

If you come across an installation that is unavailable and required for your work, please let us know by sending a request to helpdesk@chpc.utah.edu.