Download Latest Version gpfsmonitor-2.2.tgz (157.8 kB)
Email in envelope

Get an email when there's a new version of GPFS Monitor Suite

Home / gpfsmonitor-2.2 / dbqueries
Name Modified Size InfoDownloads / Week
Parent folder
host-info 2015-01-20 1.8 kB
0
L 2015-01-20 1.6 kB
0
latest-entries 2015-01-20 2.6 kB
0
latest-fs-status 2015-01-20 3.3 kB
0
latest-waiters-by-host 2015-01-20 2.5 kB
0
q 2015-01-20 2.5 kB
0
q2 2015-01-20 2.0 kB
0
q3 2015-01-20 2.4 kB
0
rdma-disabled 2015-01-20 1.8 kB
0
vmm-waiters 2015-01-20 2.2 kB
0
waiter-reason-counts 2015-01-20 2.1 kB
0
latest-errors 2015-01-20 3.0 kB
0
NOTES 2015-01-20 2.6 kB
0
q1 2015-01-20 2.0 kB
0
waiter-reasons 2015-01-20 1.7 kB
0
Totals: 15 Items   34.2 kB 0 
GPFS Monitor Suite for Linux
----------------------------

The GPFS Monitor Suite is a set of perl script daemons that forward snapshots of
a node's GPFS state to a central monitor and reporter, which in turn provide a
view of the state of nodes and file systems of one or more GPFS clusters.
The monitor state information is gathered by parsing the output of several GPFS CLI
programs along with testing file system accessiblity via the POSIX API.  It has been
found advantageous to perform this monitoring as a user task, as there are a number of
scenarios where the GPFS mmfsd daemon may be somewhat responsive to GPFS management node
probes, but the node is still unusable for end-user work.

This suite is in use at the National Center for Atmospheric Research (NCAR) to monitor
in the three GPFS storage environments located at the NCAR Wyoming Supercomputing Center
(NWSC).  At the start of January 2015, the largest GPFS multicluster based storage subsystem,
named GLADE, consists of approximately 16 PB of RAID storage, 20 NSD nodes and 6 management
nodes, with a total of 4,744 GPFS client nodes.  The largest single component is the
Yellowstone compute cluster, which consists of 4,536 GPFS client compute nodes.
Source: README, updated 2015-01-20