CLE 7 Upgrade 2021
Please note that raad2 is now available for your computational work again. As we shared in our previous update, this recent maintenance window was used for extensive software upgrades on the system. Because a lot has changed “under the hood”, these changes will require some adjustments from the users.
Application Re-Compilation
By far, the single largest adjustment is the need to re-compile previously installed applications. Because the base operating system and therefore the core system libraries have been updated, applications compiled against the older versions of those libraries may or may not work correctly (or at all) in the updated environment. Problems exhibited by pre-update executables will not be investigated by RC going forward. If you wish to use a particular application on the system, it must be re-compiled from scratch at this time. This is an unpleasant side-effect of such upgrades, but is unfortunately unavoidable.
RC is already working on re-compiling applications that are (based on the pre-upgrade user survey) employed by a multitude of users. We will gradually assist all users to re-compile all necessary applications. At the same time, we invite those users who are capable of doing so to re-compile their own applications in order to speed up a return to productive work. Some of you have in fact compiled and installed your own instances of relevant applications in your home directories in the past.
RC Installed Applications
Applications that RC staff compile and make available in system-wide locations are typically accompanied by associated module files. As users have done in the past, they will continue to use the module command to set up the appropriate environment variables before invoking their respective applications. Please note that the old set of module files have been removed and a new set is being created as new instances of applications are compiled. Newly compiled applications shall be placed under the new directory /lustre/sw/xc40/cle7 while their associated module files will be found under /lustre/sw/xc40/cle7/modulefiles. While the syntax of the module commands remains unchanged, the names of various module files will have changed. References to module files in your slurm job scripts must be adjusted accordingly. If you have explicitly referenced paths to your application binaries in slurm jobs scripts in the past, those paths must also be updated accordingly.
To see which applications are currently available on the system, use the module avail command as always. For a more concise list of TAMUQ installed applications (as opposed to Cray supplied software), you can issue the tamuq_software command. This list will keep growing over the coming days.
Anaconda Virtual Environments
There is another class of users who rely on python codes, and the use of anaconda virtual environments to create their workflows and application frameworks. These virtual environments too must be re-created. From the results of the user survey mentioned earlier, most of these users are able to do this work on their own. Some updated guidance on creating anaconda virtual environments can be found here:
Reporting Potential Problems
Because the changes on the system are quite extensive, it is inevitable that we will find some things out of order in the coming days as users return to the system and start kicking the tires and doing some test driving. We would appreciate your descriptive feedback on such issues if and when you come across them. We would also appreciate your continued patience as we work through those issues once they have been reported. As always, the best reporting mechanism is via the service desk (servicedesk@qatar.tamu.edu). Relevant and descriptive subject headers and complete and detailed descriptions of the problem are always helpful and appreciated.