System Announcements

Oct 10, 2019: 
Power is back! The power feeds were restored to the Lab this afternoon. LBNL Facilities will do high voltage switching tonight and then a build-by-building re-energization tomorrow. Our building is currently prioritized towards the end and datacenter is completely down so it'll be another 12-24 hours for our datacenter group to get the environment stabilized. Assuming everything goes well, we will start bring up the parallel and NAS filesystems and some infrastructure Sunday evening and then the compute nodes starting first thing Monday morning. Our plan is to have Lawrencium and other HPCS systems back online sometime Monday afternoon.


Oct 08, 2019: 
We are anticipating that PG&E power to the Lab will be shut off Wednesday morning (October 9) at 12:01am due to the Fire Weather Watch and likelihood of high winds.

Based on this information, we are planning to shutdown Lawrencium and all other clusters at 10pm in advance of the power outage. Note that this outage will affect access to the Lawrencium Supercluster, SVM (Scientific Virtual Machine) service, Software Modules Farm, and any systems in the 50B-1275 and 50A-2109 data centers.

We apologize for the inconvenience. Users can expect these systems to be back into production about 8 hrs after power is restored.

Thanks for your cooperation


June 18, 2019:
We experienced issues with the slurm database today causing slurm to be down from 08:59 until 17:30.   We are now back in full production.  If you experience any issues please open a SN ticket by sending an email  to hpcshelp@lbl.gov.

Now in Production:

LR6 partition: 72 nodes with Intel Xeon Skylake processors. Each node has 32 cores and FDR infiniband, Among those nodes, 48 have 96 GB memory and 24 nodes have larger memory of total of 192GB.

CM1 partition: 14 node with AMD EPYC processors. Each node has 48 cores, 256 GB of memory and FDR infiniband.

ES1 GPU partition: 24 nodes in total. 12 nodes each equipped with 4 NVIDIA  GTX1080TI cards and another 12 nodes each quipped with 2 NVIDIA V100 cards with NVLink.

Scheduled Shutdowns and Downtimes


Sep 19, 2019 : 

The Lawrencium Supercluster is scheduled a downtime due to power work (datacenter 1275) by LBLNet, starting 9/19/2019 Thursday 5pm to 9/23/2019 Monday

May 21, 2019 : 

The Lawrencium Supercluster maintenance is complete as of 17:00 5/21/2019.


May 13, 2019 : 

The Mako cluster has been shutdown and decommissioned. 

Lawrencium Clusters

LR6


 

LR5

LR4


LR3

LR2



ES1

CM1