ARC-TS Winter 2021 Maintenance

Due to maintenance, the high-performance computing (HPC) clusters and their storage systems  (/home and /scratch) will be unavailable:

  • Great Lakes: Monday, January 11, 2021, 6 a.m. – 6 p.m.
  • Armis2, Lighthouse: Monday, January 11, 2021, 6 a.m., to Tuesday, January 12, 2021, 6 p.m.
  • ConFlux: Monday, January 11, 2021, 6 a.m., to Tuesday, January 12, 2021, 6 p.m.
  • Cavium ThunderX: Monday, January 11, 2021, 6 a.m., to Wednesday, January 13, 2021, 6 p.m.

Copy any files you might need during the maintenance window to your local drive using Globus File Transfer.

Contact arcts-support@umich.edu if you have any questions.

Planned Updates:

Great Lakes:

  • CentOS 7.9 OS update
  • Slurm 20.11 (Slurm version will remain at 20.2 until further testing)
  • NVIDIA driver updates
  • OFED updates
  • InfiniBand updates
  • Open OnDemand 1.8 (potentially to 2.0)
  • GPFS updates

Lighthouse:

  • CentOS 7.9 OS update
  • Slurm 20.11 (Slurm version will remain at 20.2 until further testing)
  • NVIDIA driver updates
  • OFED updates
  • Open OnDemand 1.8 (potentially to 2.0)
  • InfiniBand updates to gateways
  • Preventative maintenance to the data center (cluster will be powered off)

Armis2:

  • CentOS 7.9 OS update
  • Slurm 20.11 (Slurm version will remain at 20.2 until further testing)
  • NVIDIA driver updates
  • OFED updates
  • InfiniBand updates to gateways
  • Preventative maintenance to the data center (cluster will be powered off)

ConFlux:

  • OS updates
  • Nvidia Driver updates
  • Cuda update
  • OFED update

ThunderX:

  • CentOS 7.9 (user software version but not kernel)
  • Performance and file system tuning
  • Yarn queue restructuring

 

Completed Changes:

Great Lakes, Lighthouse, and Armis2:

NEW version (after maintenance) OLD version (before maintenance)
CentOS 7.9
  • kernel 3.10.0-1160.6.1.el7
  • glibc 2.17-317.el7
  • ucx 1.9.0-1.51258
CentOS 7.8
  • kernel 3.10.0-1127.18.2.el7
  • glibc-2.17-307.el7
  • ucx 1.8.0-1.50218
Mlnx-ofa_kernel-modules 
  • 5.1-OFED.5.1.2.5.8.1 (Great Lakes only)
  • Kver.3.10.0_1160.6.1.el7.x86_64
  • 4.9-OFED.4.9-2.2.4.0 LTS(Armis2 and Lighthouse only)
  • kver.3.10.0_1160.6.1.el7.x86_64
Mlnx-ofa_kernel-modules 
  • 5.0-OFED.5.0.2.1.8.1
    • kver.3.10.0_1127.18.2.el7.x86_64
Slurm 20.02.6 compiles with:
    • PMIx
      • /opt/pmix/2.2.4
  • /opt/pmix/3.2.1
    • hwloc 1.11.8-4.el7 (OS provided)
  • ucx 1.9.0-1.51258 (OS provided)
  • slurmrestd
Slurm 20.02.6 compiles with:
  • PMIx
    • /opt/pmix/2.2.4
    • /opt/pmix/3.1.5
  • hwloc 1.11.8-4.el7 (OS provided)
  • ucx 1.8.0-1.50218 (OS provided)
  • slurmrestd
PMIx LD config /opt/pmix/2.2.4/lib LD config PMIx /opt/pmix/2.2.4/lib
PMIx versions available in /opt :
    • 1.2.5
    • 2.1.3
    • 2.2.4
    • 3.1.5
  • 3.2.1
PMIx versions available in /opt :
  • 1.2.5
  • 2.1.3
  • 2.2.4
  • 3.1.5
hwloc versions available in /opt:
    • 2.1.0
  • 2.2.0
hwloc versions available in /opt:
  • 2.1.0
NVIDIA driver 450.80.02 NVIDIA driver 450.51.05
GPFS client 4.2.3.24  GPFS client 4.2.3-22
Open OnDemand 1.8.18-1.el7 ( available on Great Lakes and Lighthouse only) Open OnDemand 1.7.14-1.el7