FastX, from StarNet, is a utility that can be used to interact with remote linux systems graphically in much more efficient and effective way than simple X forwarding. A FastX interactive session can be disconnected from without closing the session, allowing users to leave applications running and resume their sessions from other locations. More information on this utility can be found on the StarNet website: https://www.starnet.com/fastx
IMPORTANT NOTICE -- May 15, 2015: CHPC now has the latest FastX2. Over the next month, users should start to migrate to FastX2, by using this newer version for new sessions. The connections you had defined with FastX will be automatically moved to FastX2, however, your active/paused sessions started with FastX will not show when you use FastX2, and vica versa. See the new FastX2 page for details.
CHPC users can download the installation client by using this link:
You will need to login to the CHPC website in order to be able to download the software. The licensing allows users to install the client on both university and personal machines. FastX clients are available for windows (msi or exe), mac (dmg) and linux (both 32 and 64 bit as tar.gz files) desktops. The licensing information is also provided in a file called licenseinfo.txt found alongside the installers.
The licensing also allows for multiple FastX server installations; please contact
CHPC if you are interested in an installation of the server on CHPC supported resources.
CHPC has the FastX server installed on all cluster interactive nodes, as well as a set of nodes allocated specifically for FastX use. These nodes are the friscopeaks nodes. They can be accessed by ssh or through the FastX interface via friscoX.chpc.utah,edu, where X is 1 through 6. The nodes frisco5 and frisco6 have graphics cards to enable the use of VirtualGL for accelerated graphics. See step 9 of the 'To Use' section below for further instruction on the use of VirtualGL.
If you have any issues with FastX, please contact User Services at firstname.lastname@example.org.
For the server the latest version is 1.0-39. The client version needs to be larger or equal than the server version, we generally supply the same client version as the server in the download page. An already installed client may ask for auto update, which is generally safe to do but please watch CHPC e-mails for cases when it's not. If you get such an e-mail or FastX stops working after the client upgrade, please, downgrade to the client version that is at the CHPC sofware download web page and wait till CHPC updates the server before updating your client.
CHPC Usage Policy:
- Users should by default use the frisco resources for FastX usage.
- The CHPC usage policy for the general interactive nodes still applies when using FastX. See Policy 2.1.1. When in doubt about the computational requirements of a workflow, it is better to use one of the frisco nodes or an private owner interactive node.
- When using FastX, users should connect directly to a specific resource. The generic pattern "cluster.chpc.utah.edu" is not the name of a single server, but a round-robin address which does not guarantee a connection to a specific interactive node. Users will need to explicitly select the address of the desired node. For ember, kingspeak, and telluride, this will be cluster1.chpc.utah.edu or cluster2.chpc.utah.edu (for example, kingspeak1 is kingspeak1.chpc.utah.edu); for ash-guest, this is ash5.chpc.utah.edu or ash6.chpc.utah.edu. Frisco peaks are frisco1.chpc.utah.edu through frisco6.chpc.utah.edu.
- CHPC may establish additional usage policies for FastX depending on evolving user needs and usage patterns.
- Start the FastX client on your desktop (Note: If you are connecting using a VPN, be sure to start the VPN prior to opening the FastX client.)
- Use the green "+" to define the server. You need to provide a name and a host (choosing from the list above). As an example, you can use the name frisco1, and the host frisco1.chpc.utah.edu
- Once you have a server defined you can select one of the servers from the list and login. When you do so the text "Start a new session..." will appear in the right hand side of your FastX window.
- Right click on this text, choose start and one of the three options (Gnome, KDE, Xterm).
The first two give you a full desktop environment, and the last just a Xterm. Your
choice depends on you needs, but for many options the Xterm is best. When you do this
the session becomes defined in the FastX window. You can also specify a more lightweight
desktop environment that is available by selecting the New option in this menu, clicking
the 'desktop' button and entering
/usr/bin/xfce4-session. Another option is to use
/usr/bin/terminalwhich will give you a xterm with more features than the plain Xterm option.
- With the Xterm option, you may need to set the environment variable (or load the module) needed for the application you want to run, and then start the application.
- If you finish, you can close this window, and terminate the session (right click on the session and select terminate). You can close the FastX window at this point. Please do this for any session that you are done to ensure there are sufficient resources available to meet the active user needs.
- If you want to keep the session in order to return to it, right click on the session and select disconnect and then also disconnect from the server. You can close the FastX window at this point.
- To reconnect, either from the same desktop or another, start FastX, select and login to the same server, and you should see your saved session. Right click on the session, and select reconnect. All windows you had open when you disconnected will appear.
- For applications that need virtualgl, you need to use servers frisco5 or frisco6. Append the starting of your application in the FastX session with
/usr/bin/vglrun. As an example to start vmd with virtualgl, you would use "
vglrun vmd". Note that in the case that your application call is a script, as is the case with using gv to start gaussview, this will not work; instead you need to add this vglrun internal to the script.
- Please be sure to close any sessions you no longer need to keep the number of sessions at a minimum.
- For users of Gaussview you will need to source /uufs/chpc.utah.edu/sys/pkg/gaussian09/EM64T/etc/g09.csh before starting Gaussview with gv. With the use of modules you would do a module load gaussian09 (unless loaded via the custom.csh script.
For users of the CSDS program mercury, you will need to use VirtualGL for the molecules to appear in the display window.
- To cut/paste to or from the FastX window: see http://www.starnet.com/xwin32kb/copy_and_paste_in_xterm
- For users of Abaqus, you must use the -mesa flag (abaqus cae -mesa) or your model will not display.