Connecting to the LinCompute Cluster

 

Audience

All users of the LinCompute Cluster.

Introduction

The LinCompute (LinuxCompute) Cluster provides UT Austin students, faculty, and staff access to a set of Linux remote servers and software. Permission does not need to be granted to access the server; anyone with an active UT EID will be able to connect.

The LinCompute Cluster consists of two nodes but is accessed as if it was a single server. The server name to connect to is lincompute.cns.utexas.edu.

Note that the LinCompute Cluster shares the same user profile disk space as the WinCompute cluster (Stat Apps Server). This makes sharing files between LinCompute and WinCompute easy. Note that there is a default quota of 20GB per user. Also accounts which are no longer used are subject to be deleted.

Prerequisites

To connect via SSH

  • Computer connected to the utexas WiFi, UT wired network, or UT VPN.

    • If you’re off campus, connect to the UT VPN using the Cisco AnyConnect Secure Mobility Client application. For instructions on getting connected, please see the UT Service Desk’s documentation: UT VPN Client: Getting Started

  • SSH Client

    • Linux & macOS can use the built-in Terminal application.

    • Windows will need to install an ssh client, we recommend PuTTY which can be downloaded from the PuTTY website. If your computer was set up by CNS IT, contact CNS OIT by submitting a request form at https://cns.utexas.edu/help or sending an email to help@cns.utexas.edu.

To connect via web browser

  • Computer with an internet connection.

  • Web browser.

Software Included

Guidelines for using the LinCompute Cluster

  • Do not store critical files on your LinCompute Cluster profile, as they may not be saved long-term and are not backed up or archived.  You can download files from UT Box, an e-mail account, or other sources to access on the server.

  • The LinCompute Cluster does not meet the requirements for storing Category-I (Confidential) data.  Do not store Category-I data on your profile on the cluster.

  • It is a shared resource; please be aware of how your use may impact others. Limit the amount of CPU and RAM you use when possible.

  • The LinCompute Cluster is intended for short-duration personal projects or for trying out a new software package prior to purchasing or running elsewhere (e.g. TACC).  We do not recommend using the LinCompute Cluster for work related to dissertations, theses, publications, or time-sensitive projects. We cannot guarantee the performance of the LinCompute Cluster at any given time due to its shared nature.

Getting Connected via SSH

Connecting for the first time

The first time that you connect to the server using the instructions below, a message will ask you if you want to continue connecting. This message appears because the remote server has an ssh key that's not stored in your system registry and as a result, the identity of the key can't be verified. This is normal for your very first connection, and should not appear when you later connect from the same machine. If this happens, simply type yes and press Enter. This action will add the ssh key to the list of known hosts on your machine so that you don't see this warning again when you make future connections from the same machine. You can expect to see that warning again, however, if you connect from a different computer.

Mac

  1. Go to Applications > Utilities, and then open Terminal.

  2. Establish an SSH connection to the server by using following command, where “your-eid” is replaced by your actual UT EID username: ssh -X your-eid@lincompute.cns.utexas.edu

  3. If you receive a message asking if you want to continue connecting, type yes and press Enter.

  4. You will be prompted for your password. Enter your UT EID password and press Enter. Note that the password does not echo to the screen as you enter it.

  5. You're done! You should now be connected to the LinCompute cluster.

Windows

  1. Launch the PuTTY application.

    1. Note: you can use your ssh client of choice, but we’ll be referencing PuTTY specifically.

  2. Use the following steps to configure your connection:

    1. In the PuTTY Configuration window, enter the following values:

      • In the Host Name field, enter the name lincompute.cns.utexas.edu

      • Ensure that the connection type is set to SSH.

      • (Optional but recommended) In the Saved Sessions field, assign a name for
        this connection (such as LinCompute). Assigning a name saves time the next time that you use Putty.

    2. Click Open.

    3. If you receive a message asking if you want to continue connecting, type yes and press Enter.

  3. You will be prompted for your username and password. User your UT EID username, and your UT EID password. Note that the password at is not echoed to the screen. Then, press Enter.

    If you have entered the correct username and password, it will log you in and respond with
    a shell prompt.

  4. You're done! You should now be connected to the LinCompute cluster.

Linux

  1. Open a terminal window.

  2. Establish an SSH connection to the server by using following command, where “your-eid” is replaced by your actual UT EID username: ssh -X your-eid@lincompute.cns.utexas.edu

  3. If you receive a message asking if you want to continue connecting, type yes and press Enter.

  4. You will be prompted for your password. Enter your UT EID password and press Enter. Note that the password does not echo to the screen as you enter it.

  5. You're done! You should now be connected to the LinCompute cluster.

Getting Connected via the Web

You can connect to a web interface via any normal web browser. Currently only R Studio is available via the web interface, but we hope to add Jupyterhub notebook soon.

Note: From the R Studio web interface you can get a Linux command line shell and execute command similar to if you used ssh to login (though without any X11 forwarding for graphic applications).

To connect via the web, you must connect to https://lincompute.cns.utexas.edu/ and select the application you want (R Studio for example). The application will ask you to login using your UT EID username and password.

Using Software

Anaconda3

The system has an Anaconda3 python environment installed. To activate it, assuming the default shell of bash is in use, enter the following command.

eval "$(/opt/anaconda3/bin/conda shell.bash hook)"

If you are using another shell other than bash, simply replace “shell.bash” with the appropriate string such as “shell.tcsh”.

Julia

Julia should be available system wide using the /usr/bin/julia (or just “julia”). You shouldn’t have to do anything special to use it.

DS9

DS9 is available at /usr/local/bin/ds9 (you may have to specify the entire path if /usr/local/bin is not in your path). You will need a functioning X11 connection to use DS9 (see the FAQ for more information on X11).

TOPCAT

TOPCAT is available at /usr/local/bin/topcat (you may have to specify the entire path if /usr/local/bin is not in your path). You will need a functioning X11 connection to use TOPCAT (see the FAQ for more information on X11).

JDK (versions 1.8.0 and 21)

By default, the system provides JDK version 1.8.0. This is the system support JDK software. JDK 21 is also installed, and can be accessed manually. To use JDK 21, use the following commands (for bash shell, modify as needed for other shells).

export JAVA_HOME=/usr/lib/jvm/java-21-openjdk-21.0.3.0.9-1.el8.x86_64 export PATH=/usr/lib/jvm/java-21-openjdk-21.0.3.0.9-1.el8.x86_64/bin:$PATH

MESA SDK

The Mesa SDK is installed at /opt/mesasdk/. For information on using it, please visit https://www.astro.wisc.edu/~townsend/static.php?ref=mesasdk

Mathematica

Mathematica is available system wide. You will need a functioning X11 connection to use many of the default mathematica interfaces (see the FAQ for more information on X11). You can run mathematica without X11 by invoking it as “math” instead of “mathematica”.

Matlab

Matlab is available system wide. A warning that it can be very slow to startup and/or use. You will need a functioning X11 connection to use many of the default matlab interfaces (see the FAQ for more information on X11). You can run matlab without X11 by invoking it as “matlab -nodisplay -nosplash” (you might also want to use -nojvm if the java vm isn’t needed, as this will speed up the startup).

Matlab has not yet been verified to work in the various configurations. It is also notoriously slow remotely. Attempt use at your own peril.

Alternatively, Matlab is available to students, faculty, and staff and can be installed on personal and UT computers. For instructions to sign up & install and to request assistance, please see this page from the UT Service Desk: MATLAB

R and R Studio

R is available system wide. It is also available via R Studio by using a web browser and going to https://lincompute.cns.utexas.edu/ (login with your UT EID credentials).

FAQs

Q: Why do I get "Permission denied, please try again" or other error trying to login to the cluster via ssh, or an application via the web?

A: This is usually due to providing the wrong user name or password, or providing the username in the wrong format. Please enter just your UT EID, without any domain information. Also, try entering your UT EID username in all lowercase.

Q: Why can’t I ssh to the cluster from off campus or from Dell Medical School?

A: Per a directive from the UT Information Security Office (ISO), ssh access using passwords from outside the UT campus network has now been blocked unless using the UT VPN service. Networks at Dell Medical School are not part of the UT campus network, so also require use of the UT VPN service. You can access LinCompute from off-campus by either using the UT VPN service, or setting up public key authentication.

Q: How can I transfer files in or out of the cluster?

A: You can use the ssh-based tools scp or sftp to transfer files between machines. You can also use a web browser running on the LinCompute machines to download or upload files from web-based services such as UT Box. Please note that WinCompute and LinCompute share the same storage so you can easily access files between those two systems without any transfer needed.

Q: Why can’t I run a GUI (Graphical User Interface) application when I login via ssh?

A: X-Windows is the display rendering (windowing) system used on Linux/Unix computers. X11 is the protocol used by X-Windows; it specifies how an application window is to be rendered (e.g. a series of lines, boxes, polygons, colors, etc, etc). The application creates a series of such steps according to the X11 protocol specification. The receiver (the X11 client) reads and follows those steps to render an image in its local environment.

The terms X11 and X-Windows are sometimes used interchangeably, but technically X11 is the protocol and X-Windows is a term for applications that use that protocol to render their windows, and the entirety of the components involved.

The important thing about X-windows is that it is a portable system. The application that uses X-windows to render its screens can run on any operating system, not just Linux/Unix. As long as the client (e.g. the Mac Terminal, or an appropriate Windows program) supports X11, the visualizations can be rendered locally. Contrast this with Windows-only or Mac-only applications which use operating system specific API calls. 

However, to run a GUI program installed on LinCompute, you'll need an X11-enabled terminal (see below).

A local terminal enabled for X-Windows (X11) is needed in order to invoke Linux GUI (graphical user interface) program and see the GUI on your laptop or desktop running Windows, Mac OS, or even Linux. This may require installing a 3rd party X-Windows package (such as Xming on Windows or XQuartz on Mac) then using an X-Windows-capable terminal (PuTTY on Windows, or the built-in Terminal on Mac or Linux) to ssh to the LinCompute Cluster.

If you invoke ssh explicitly from a Mac Terminal session or Linux terminal session, use either -X or -Y to connect to the LinCompute Cluster; the -X or -Y enables forwarding of the X11 commands to the X-terminal). You might also want to use -C (compression) which will generally provide a slightly faster and more responsive session. On Windows, you configure PuTTY to use X11 forwarding automatically when you open a session.

Q: Why are remote GUI (X11) sessions so slow?

A: X11 is not a very efficient protocol, and your remote network speed can greatly effect the rendering speed of GUI applications. Using VPN may slow it down even more. If you need high performance GUI access, it is recommended that you connect from on-campus rather than off-campus.

If you must connect from off campus via ssh, it is recommended you use the -C (compression) option for ssh, which will generally provide a slightly faster and more responsive session than without using compression.

Q: When running ipython, I get an error (“sqlite3.OperationalError: database is locked”)

A: We are looking into a fix for this issue, but until then, you can use the following workaround.

mkdir /tmp/${USER} chmod go-rx /tmp/${USER} ipython --HistoryManager.hist_file=/tmp/${USER}/ipython_hist.sqlite

You will need to use the last line every time you invoke ipython (you can set an alias for it if desired).
Another option is to add the following line to your .bashrc file to tell ipython to put all your files in the above location:

This isn’t quite as safe since it not only relocates your history file, but all ipython data files — configuration files, command history and extensions—in this directory.

Note that in either of these cases, the files in /tmp will be local to the lincompute node you are on, and not shared between nodes like your home directory is.

Still have questions or need assistance?

Contact CNS OIT by submitting a https://cns.utexas.edu/help or sending an email to help@cns.utexas.edu