User Tools

Site Tools


gpu:file_systems

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Next revision
Previous revision
gpu:file_systems [2010/10/08 17:27]
mlawrence created
gpu:file_systems [2021/12/09 16:42] (current)
Line 1: Line 1:
-The GPU cluster is provisioned with two filesystems mounted on all the GPU compute nodes as follows : 
  
-Filesystem            Size  Used Avail Use% Mounted on +The GPU cluster is provisioned with a set of two shared filesystems mounted on the head node as well as on all the GPU compute nodes as follows :
-/dev/sda1              48G  8.0G   38G  18% / +
-/dev/sda2              48G  180M   45G   1% /spare +
-tmpfs                  12G       12G   0% /dev/shm +
-/dev/gpfs              12T   45G   12T   1% /GPU +
-sn8-ib@o2ib:/lustre   7.9T  5.7G  7.5T   1% /LUSTRE+
  
 +  Filesystem            Size  Used Avail Use% Mounted on 
 +  /dev/gpfs              12T   45G   12T   1% /GPU  
 +  sn8-ib@o2ib:/lustre   7.9T  5.7G  7.5T   1% /LUSTRE 
  
-The "/GPU" filesystem of 12TB is served by a set of four (4) GPFS NSD Servers in the storage cluster. This mount is shared between all the nodes (hn, cn1, cn2,cn3,cn4,cn5). System administrators should refer to the Wiki namespace on GPFS for configuration and operations procedures.+The **"/GPU**" filesystem of 12TB is served by a set of four (4) GPFS NSD Servers in a GPFS storage cluster. This mount is shared between all the nodes (hn, cn1, cn2,cn3,cn4,cn5). System administrators should refer to the seperate Wiki namespace on GPFS for configuration, management and operations of the GPFS cluster.
  
-The "/LUSTRE" filesystem mount is a 7.9TB shared filesystem served by a Lustre cluster running on a single shared Lustre MDS/MGS, plus a set of 3 ODS servers, serving a total of 9 Lustre OST's (3 per ODS). System administrators can refer to the Wiki namespace on Lustre for configuration and operations prcoedures+The **"/LUSTRE**" filesystem mount is a 7.9TB shared filesystem served by a Lustre cluster running on a single shared Lustre MDS/MGS, plus a set of 3 ODS servers, serving a total of 9 Lustre OST's (3 per ODS). System administrators can refer to the Wiki namespace on Lustre for configuration, management and operations of the Lustre cluster. 
 + 
 +Connectivity between GPFS clients, Lustre clients and the respective storage cluster servers uses 40Gbps QDR Infiniband, with GPFS using Direct Memory Access (RDMA) over Infiniband. 
 + 
 + 
/app/dokuwiki/data/attic/gpu/file_systems.1286551634.txt.gz · Last modified: 2021/12/09 16:42 (external edit)