Effective July 1, 2016, the 30-day file retention policy for networked scratch filesystems on both Helium and Neon will change.
The FhGFS (/scratch) file system has been unmounted on compute nodes as discussed in the entry, Helium FhGFS scratch decommissioning | High Performance Computing.
What was done:
- Package updates for Helium
- Fix initscripts on Neon
- Expand ephemeral port range on Neon
- Rolled over SGE log files
The next scheduled maintenance window for the HPC systems is May 26, 2016 7 AM to May 27, 2016 7AM.
A reminder that long running resource intensive jobs should not be run on login nodes.
As part of the Helium decommissioning process the FhGFS scratch file system will be decommissioned from compute nodes on July 1, 2016. Note that this is the high performance file system mounted at /scratch and not the more commonly used NFS scratch file system mounted at /nfsscratch.