Skip to Main Content


If you are looking to create an account or otherwise gain access to this cluster, please see this FAQ entry.

Accessing ada via the Login Node

Your ada login credentials exactly match what you enter into your single sign-on interface to the UMBC portals. Consider your username as $USER in this document.

To access ada, use your favorite remote connection client (e.g., VSCode, Bitvise, PuTTY) or simply run ssh $ at a command line. Explore the man pages on ssh for more options.


Upon successful login, take note of the legal warning and notice, and also pay attention to the remainder of this text, called the message of the day (MoTD). The MoTD is the main way that system administrators will update users concerning changes to the cluster environment. Once you receive a prompt (e.g. something like [$USER@ada ~]$) you’ll know that the contents of your ~/.bashrc file have been run and executed successfully and your working directory will be your home directory.

The machine is the login node or “edge” node for the cluster environment. “Edge” since it straddles the edge between what machines are accessible from the larger campus network and the machines that are not accessible from this larger network (e.g. the GPU nodes).


It should be noted that the login node has minimal computational resources and code should never be run on the login node. Doing so may make it difficult for others to use the system and may result in restrictions to your account.

Username and Group

Every user belongs to a primary group that exists within the cluster environment. With $USER representing your UMBC username that exists as your user identifier within the cluster environment, $GROUP will represent this primary group to which you belong. For many users, their primary group is the group created for their sponsoring faculty member. It is common that users belong to many groups (indicating collaborations or simply the sharing of data), these groups are called supplemental groups.

Any user can look at the groups to which they belong by running groups at the command line once logged into ada.

Default Symbolic Links

Every new user account that is created in this environment has a few symbolic links in their home directory that allow them quick access to their research storage volumes. Think of these are shortcuts out of your home directory—which is limited in size—to the larger volumes described under “Group Storage” on the Storage page. A summary of these links is given below.

Link Name Linked To
~/${GROUP}_common /nfs/rs/$GROUP/common
~/${GROUP}_user /nfs/rs/$GROUP/users/$USER
~/ada_${GROUP} /nfs/ada/$GROUP
~/.cache /nfs/ada/$GROUP/users/$USER/.cache
~/.conda /nfs/ada/$GROUP/users/$USER/.conda

Due to their size and complexity, none of the above group storage volumes are backed up. Only home directories are backed up. It is therefore imperative you work to safeguard your work by periodically copying important results or code to your own storage.


At the same time your user account is created on ada, your account within the cluster workload manager is created. This gives you access to the compute resources available within the cluster environment. More on this is described in the SLURM pages.

In order to test your SLURM account, run the following simple test to confirm access to the compute resources.

srun --mem=1 --time=1 --gres=gpu:1 hostname

This will request 1MB of CPU RAM for 1 minute on any available GPU. Once these resources are identified and allocated to you, the small executable hostname will run on that hardware. This srun command executes this hostname command on this remote system and redirects the stdout to the stdout on the login node. So an example of this test might look like the following.

[$USER@ada ~]$ srun --mem=1 --time=1 --gres=gpu:1 hostname
[$USER@ada ~]$

Please note that if the cluster is under heavy utilization, this may take a while to return.


If you notice any issue with this SLURM submission or any discrepancy with your user environment and the default described on this page, please send an email to or submit a descriptive help ticket via the Research Computing Help Request System (see the Forms drop-down above).