Pitzer

System Downtime January 7, 2020

A downtime for all HPC systems is scheduled from 7 a.m. to 5 p.m., Tuesday, January 7, 2020. The downtime will affect the Pitzer, Ruby and Owens Clusters, web portals, and HPC file servers. System access to storage and license servers hosted by OSC will not be available during this time. Login services, except for my.osc.edu, will not be available during this time.

OnDemand Version Update

On Wednesday, December 11 from 9am-915am there will be a short interruption of service for OnDemand to deploy a new version of the portal. It will deploy some updates to the web apps and infrastructure. **During this interruption any active shell connections, noVNC connections, and file transfers will be terminated. Afterwards one should be able to reconnect to running VDI and iHPC sessions.** Contact oschelp@osc.edu if there are any questions.

GPFS filesystem Problem

We are having issues with GPFS filesystem (both project and scratch) on all clusters. It started around 3:08 PM, Oct 24, 2019. This also causes the inaccessibility of OSC OnDemand. We are not clear whether there are any job failures due to this problem. We are working to resolve the issue now. We will keep you posted and apologize for any inconvenience. Please contact oschelp@osc.edu if you have any questions.

Ls-dyna license outage

The ls-dyna license server that OSC is hosting has not been available since Oct 15, 2019. This outage is due to hardware failure on the server. We apologize for any inconvenience this may cause you. Please contact oschelp@osc.edu if you have any questions.

TotalView Debugger Removal

The TotalView license server will shutdown and support of TotalView installed on OSC systems discontinued, along with the statewide licensing, on Nov. 1, 2019. Recently, a security vulnerability was found on the old license server software. OSC has a permanent Totalview license, but it is without the maintenance of the software. Thus, we cannot update the software to a new version. We have provided another HPC debugger, DDT, from ARM. Please consider migrating to DDT. Contact OSC Help for comments, questions or concerns.

Scratch Filesystem Errors Resolved

We have resolved a problem with the Scratch filesystem that gives errors "No space left on device". The cause is that the metadata subsystem is pretty full. The outage was caused be the rapid exhaustion of metadata storage space, the reason for the spike in metadata space utilization is unknown. The scratch filesystem has been returned to service after we worked with the vendor to implement a temporary workaround. We are working with the vendor to find the root cause and a permanent fix for the underlying problem. We apologize for any inconvenience this may cause.

Pages