NIH HPC News & Announcements
EXTENDED DOWNTIME for Biowulf/HPC systems June 22-25, 2023
Date: 31 May 2023 09:05:56
From: NIH HPC Systems Staff
There will be an extended downtime of the Biowulf cluster starting on Thursday June 22 at 6 pm and running until Sunday June 25 at 10 pm to allow for the Biowulf cluster, the Biowulf login node, and Helix system to be transitioned from the RHEL7/CentOS7 operating system to the RHEL8/Rocky8 operating system. Following the downtime, the majority of cluster nodes will be available for jobs, but there will be fewer resources for a few days as remaining nodes are transitioned to RHEL/Rocky8.
During this extended maintenance window the following HPC services will be unavailable to all Biowulf/HPC users:
- Biowulf login node & cluster
- Helix (the HPC data transfer node)
- HPCdrive
- the NIH HPC Globus endpoint
- Partek Flow
Note that a batch system reservation will be put in place for the downtime period. This means that the batch system scheduler will not start jobs that wouldn't end before the start of the reservation. Therefore, you should carefully choose job time limits so as not to unnecessarily delay your jobs. Any running jobs will be terminated on June 22 at 6 pm.
For most users, this transition will not affect their jobs or workflow. Most applications have been rebuilt for Rocky8 and will be available just as they are now. However, there are a few changes that users need to be aware of:
* Application installations and updates on the RHEL7/CentOS7 system have been frozen until the transition. The HPC staff is currently migrating several hundred applications to Rocky8, so no new apps will be installed and no applications will be updated on the current CentOS7 system.
* Only recent version(s) of each application will be rebuilt for Rocky8. If your scripts specify a version of an application, you may need to update your scripts. If you absolutely require an older version of an application, let us know.
* User-built applications: Applications that were compiled by users under CentOS7 _may_ need to be recompiled for new libraries. CPU-performance sensitive applications (e.g. molecular simulations) should be recompiled. All Biowulf/HPC users now have access to the RHEL8/Rocky8 cluster per email sent on May 30 in order to start such work.
Details about the changes are listed at https://hpc.nih.gov/docs/rhel8.html
Questions? Send email to staff@hpc.nih.gov
HPC Staff.
########################################################################
Please contact staff@hpc.nih.gov with any questions about the NIH HPC Systems
[Last 12 months of HPC announcements]