File Management

Nightingale Filesystem

Nightingale uses a Lustre shared filesystem on all of its nodes. Lustre is an object-based, parallel distributed filesystem for large-scale, high-performance computing clusters. It enables scaling to a large number of nodes (tens of thousands), petabytes (PB) of storage, and high aggregate throughput (hundreds of gigabytes per second). These features make Lustre an advantageous data storage solution for many scientific computing applications.

The table below describes the location, purpose, and quota policy for the storage areas available on Nightingale’s filesystem. More information about these storage areas is described in the following sections.

Nightingale Storage Areas
Location	Purpose	Quota Policy
/u	home	50GB
/projects	projects	1TB (more can be purchased)
/scratch/users	user scratch
/tmp	local scratch
/datasets	datasets

Home

The /u area of the filesystem is where users land upon logging on to the cluster and is where a user’s $HOME environment variable points. This area has a fairly small quota and is meant to contain a user’s configuration files, job output/error files, and smaller software installations. This area is automatically set up during the account provisioning process and there is no additional charge for this storage.

It is not possible to request an expansion of the home directory quota. If a user depletes the available space on their home directory, they will be notified and given the opportunity to remove files from it. You will need to delete files to get below the threshold and will not be able to store additional data until you do so.

Projects

The /projects area of the filesystem is where group members (whether a single faculty member, a lab group, a department, or an entire college) store their project-related files. A user can have access to multiple project subdirectories if they are a member of multiple groups and have been granted access to the space by the project’s Principal Investigator (PI).

Projects pay a separate charge for this project space and the minimum allocation is 1 TB. Additional space can be purchased as needed.

User Scratch

The /scratch area of the filesystem is where users can place data temporarily while a running job is using it.

Local Scratch

The /tmp area is a local filesystem on an individual compute node and is not part of the shared file system. Data placed in /tmp is purged when the job completes.

Datasets

The /datasets area contains curated, read-only datasets typically exported from a Postgres database. Projects will be informed when data is placed here for them.

File and Directory Permissions

When you create files or directories on Nightingale, do not grant permissions to “others”; this includes read, write, or execute. You can set “owner” and “group” permissions to meet your needs. For more information on Linux file permissions, review Linux file permissions explained.

File and directory permissions on Nightingale are monitored to maintain the integrity of protected data on the system. If you grant file permissions to “others”, system admins will be alerted, and you will be contacted to investigate.

Transferring Files

Copying Project Data to Home, Scratch, or Projects

You will not work directly on your group’s curated project data in the /datasets directory. Instead, you will need to copy the data from the original directory to your home, scratch, or projects directory. Follow the steps below to copy a dataset to your home directory:

Enter the cd command to go your home directory.
```
cd
```
Use the mkdir command to create a new folder in your home directory using a name of your choosing.
```
mkdir <folder_name>
```
Use the cp command to copy the dataset to the folder you created. The cp -R command is used for recursive copy of all files and directories in the source directory tree.
```
cp -R /<dataset_foldername>/<filename> ~/<your_data_folder_name>
```
For example, if the dataset you want is located in ‘/datasets/covid_1’ and you want to move it to your directory ‘my_covid_data,’ you would enter the command:
```
cp -R /datasets/covid_1 ~/my_covid_data
```
To check if you copied the data successfully, enter the commands:
```
cd ~/<your_data_folder_name>/<filename>
ls
```
Using the example in the previous step, the commands would be:
```
cd ~/my_covid_data/covid_1
ls
```

Use the steps above to copy a dataset to your scratch or projects directory with the below modifications:

In step 1 above, navigate to your scratch or projects directory using the cd command.
In steps 3 and 4 above, replace the ~ with the path to your scratch or projects directory.

Copying Files onto Nightingale Using SCP

SCP (Secure Copy) is an application that gives users a secure way to copy files between machines over an unsecured network. Its syntax is similar to that of SSH used to log into a remote machine.

SCP requires a source and a destination. You can use it to copy individual files or directories. The source and destination are specified with a file path if it is on your local machine or as <login_name>@<machine_name>:<file_name> if it is on a remote machine.

Since Nightingale has a bastion host which all network traffic travels through, you need to specify that the copy will jump through the bastion. For example, a user, “test1”, copying the file “my_data” from their current directory on their local machine to their home directory on the Nightingale login node “ng-login01” would use the following command:

scp -J [email protected] my_data test1@ng-login01:.

Copying Files onto Nightingale Using AWS S3 Buckets

To use AWS S3 Buckets you must first configure the service. Run the command aws configure and answer its prompts for the following data:

AWS Access Key ID
AWS Secret Access Key
Default region name
Default output format

Copy files from the bucket using:

aws s3 cp s3://<bucket-name> <local name on nightingale>

Copying Files off of Nightingale

Any method that can transfer data onto Nightingale can also be used to transfer information off of the machine.

Before transferring data off of Nightingale, please read about protected data. Data transfers off of Nightingale are audited and must be accounted for.

Transferring Files with Globus

Use Globus for large data transfers. Globus is a web-based file transfer system that works in the background to move files between systems with Globus endpoints.

Go to Transferring Files - Globus for instructions on using Globus with NCSA computing resources.

The Nightingale endpoint collection name is “ncsa#ngale”.

Creating and Editing Files

Warning

When you create files or directories on Nightingale, do not grant permissions to “others”; this includes read, write, or execute. See File and Directory Permissions for more information.

Sometimes, it is easiest to create and edit your files directly on the cluster rather than transfer them back and forth. You can use various programs on clusters for working with plain text files; examples include vi/vim, gedit, nano, and emacs. The vi/vim text editor is one of the most commonly used. However, if you are new to working in the Linux environment, the nano editor is recommended because it may be more similar to how you edit text files on a non-Linux-based machine. Several tutorials are available online if you want to know more about nano or vi; a couple of suggestions are listed below.

You can also edit files using MobaXterm’s text editor. Brief instructions for using GNU nano and MobaXterm are given below.

GNU nano

GNU nano is an easy-to-use command line text editor for Linux. To open an existing file or create a new one, type nano followed by the file name. This opens a new editor window in your terminal where you can start editing the file.

nano file_name

At the bottom of the window, you will find a list of shortcuts to use with the nano editor. The caret symbol (^) represents the Ctrl key; for example, to exit, nano shows ^X, type Ctrl+X. The letter M represents the Alt key; for example, to undo, nano shows M-U, type Alt+U.

MobaTextEditor

If you use MobaXterm to log into Nightingale, you will see a file browser in the left pane of the MobaXterm window. Double-click on a selected file to open it in a separate window. Note that a temporary copy of files will be saved on your local machine when you use MobaTextEditor. The temporary files are saved in the AppDataRoaming folder on Windows and will be removed when you fully close MobaXterm on your machine.

Organizing Files

How you organize your files depends on how the directory structure is set up on your cluster and possibly guidelines set up by your project manager. However, there are some basic goals you should keep in mind:

File names should be logical so that you can find them a week from now, a month from now, and a year from now. Data that cannot be found later is not worth nearly as much as data you can quickly locate. When storing your files to an archive, organize them with this goal in mind so they are stored logically and can be accessed easily.
Use Tar, Winzip, or a similar file-bundling software to create a single file from a directory tree, and then store that file. The name of that bundled file should clearly indicate what files are bundled within.
In an archival storage site, apply meaningful names to files and directories so you or your colleagues can navigate back to the data when needed. It is up to you to decide what works best.