Rivanna will be down for maintenance on Monday, December 18, 2023 beginning at 6 a.m.
You may continue to submit jobs until the maintenance period begins, but if the system determines your job will not have time to finish, it will not start until Rivanna is returned to service.
All systems are expected to return to service by 6 a.m. on Tuesday, December 19.
IMPORTANT MAINTENANCE NOTES
The operating system will be upgraded to Rocky 8.7 with system glibc 2.28 and GCC 8.5.0. Due to fundamental changes in system libraries, the entire software stack is rebuilt. Users should rebuild all self-compiled codes and R packages. Starting Nov 21, users who want early access to the development environment to rebuild/test codes against the new software stack can login by running
ssh udc-ba33-36 on the frontend. (Please be sure not to overwrite your existing codes for the production environment.) Contact us here if you need assistance.
Compilers and toolchains have been consolidated to the following:
Singularity has been renamed to Apptainer. Load the
apptainer/1.2.2 module for containers. (The
singularity command is provided as an alias.) All users can now build containers directly on Rivanna; see here for details.
There are many module version upgrades and deprecation of older versions. Run
module spider NAME to check the available versions and the corresponding load command. Contact us here if you need a different version. Only the most important changes are listed below:
|OOD RStudio Server
||1.0.143, 1.1.463, 1.3.1073, 2023.03.0
||7.1.0, 9.2.0, 11.2.0
||4.6.2, 4.7.3, 4.7.4
||2.7.16, 3.6.6, 3.6.8, 3.7.7, 3.8.8
||3.5.3, 3.6.3, 4.0.3, 4.1.1, 4.2.1
- C/C++/Fortran users who must build code with GCC 7 or older should containerize the application starting with the official GCC base image. Contact us if you need assistance.
- Intel 18.0 modules are either migrated to the newer version (2023.1) or dropped. Intel users should rebuild code with
intel/2023.1 if possible.
- RStudio Server is now backed by a native module with R as a dependency. R packages installed via the
R module will be detected automatically in RStudio Server, and vice versa. All R packages will need to be rebuilt.
- Python 2.7-dependent modules are completely removed from the software stack. Users of legacy Python code can create a custom environment using the
mamba (recommended) module.
- Code Server is backed by a native module instead of a container. This allows usage of compilers and interpreters on Rivanna. Python users please see instructions here.
- Mamba is separated from
anaconda into its own module.
- Java module versions are standardized to 7, 8, 11, 12 (previously 1.7.0, etc.).