Most users will not need to run graphical applications on a remote Unix system.. If you know that you will need to use a graphical user interface with your research software, or if you want to use a graphical editor such as eclipse or emacs on over the network, read on. Otherwise, you can skip this section for now.

Unix uses a networked graphics interface called the X Window system. It is also sometimes called simply X11 for short. ( X11 is the latest major version of the system. ) X11 allows programs running on a remote Unix system to display graphics on your screen. The programs running on the remote system are called clients, and they display graphical output by sending commands (magic sequences) such as "draw a line from (x1,y1) to (x2,y2)" to the X11 server on the machine where the output is to be displayed. The computer in front of you must be running an X server process in order to display Unix graphics, regardless of whether the client programs are running on your machine or a remote machine.

Apple's macOS, while Unix compatible at the API and command-line, does not include X11 by default. It instead uses Apple's proprietary GUI. X11 for macOS is provided by the XQuartz project: https://www.xquartz.org/. XQuartz is free open source software (FOSS) that can be downloaded and installed in a few minutes. It should start automatically when either local or remote X11 clients attempt to access your display, but you can also start it manually.

Modern Unix systems such as BSD, Linux, and macOS have most of the necessary tools and configuration in place for running remote graphical applications. Some Unix servers may be configured without a GUI. If you want to remotely log into such a server and run X11 programs from your desktop or laptop system, you will need to at least install the xauth package on the remote system. This allows your system to configure X11 permissions for the remote system when you log in using ssh -X or ssh -Y.

# Debian
shell-prompt: apt install xauth

# FreeBSD
shell-prompt: pkg install xauth

# RHEL
shell-prompt: yum install xorg-x11-xauth
        

Some additional steps may be necessary on your computer to allow remote systems to access your display. This applies to all computers running an X11 server, regardless of operating system. Additional steps that may be necessary for Cygwin systems are discussed in the section called “Graphical Programs on Windows with Cygwin”.

If you want to run graphical applications on a remote computer over an ssh connection, you will need to forward your local display to the remote system. This can be done for a single ssh session by providing the -X flag:

shell-prompt: ssh -X joe@unixdev1.ceas.uwm.edu
        

This causes the ssh command to inform the remote system that X11 graphical output should be sent to your local display through the ssh connection. ( This is called SSH tunneling. )

If you want to forward X11 connections to all remote hosts for all users on the local system, you can enable X11 forwarding in your ssh_config file (usually found in /etc or /etc/ssh) by adding the following line:

ForwardX11 yes
        

Some X11 programs require additional protocol features that can pose more security risks to the client system. If you get an error message containing "Invalid MIT-MAGIC-COOKIE" when trying to run a graphical application over an ssh connection, try using the -Y flag instead of -X to open a trusted connection.

shell-prompt: ssh -Y joe@unixdev1.ceas.uwm.edu
        

You can establish trusted connections to all hosts by adding the following to your ssh_config file:

ForwardX11Trusted yes
        

If you are using ssh over a slow connection, such as home DSL/cable, and plan to use X11 programs, it can be very helpful to enable compression, which is enabled by the -C flag. Packets are then compressed before being sent over the wire and decompressed on the receiving end. This adds more CPU load on both ends, but reduces the amount of data flowing over the network and may significantly improve the responsiveness of a graphical user interface.

shell-prompt: ssh -C -X joe@unixdev1.ceas.uwm.edu
        

It is possible for Unix graphical applications on the remote Unix machine to display on a Windows machine with Cygwin, but this will require installing additional Cygwin packages and performing a few configuration steps on your computer in addition to those discussed in the section called “Configuration Steps Common to all Operating Systems”.

It is possible to run Open 3D graphical applications on remote systems as well, but performance may or may not be acceptable. There are also numerous problems that can arise that depend on the operating system and video drivers used on each end. In any case, OpenGL applications will require additional X11 components to be installed on both the remote machine running the application and the display machine running the X11 server. Determining the minimal set of packages required for every platform would be excessively difficult, so we recommend simply install the entire Xorg system on the remote system. This will likely enable at least some OpenGL applications to run remotely. This proved sufficient to run the 3D mesh visualizer MeshLab comfortably on a remote FreeBSD machine from a FreeBSD display. A similar application called VMD proved too sluggish to use remotely.

# Debian
shell-prompt: apt install xorg

# FreeBSD
shell-prompt: pkg install xorg
        

You may find it easier to simply download the data files to your local machine and run the 3D graphics applications there. Performance will be much better when using direct rendering, where the display is on the same machine running the 3D application. When using indirect rendering, where the 3D application is running on a different machine than the display, sending graphics commands over a network can be a bottleneck.