Simple NSD File

This section describes a simple NSD file with an example.

Simple NSD file can only be used for full disk that have not already been partitioned as input to Ambari.

All disks of GPFS™ NSD server nodes requires to be listed in the NSD stanza file.

The following is an example of a preferred simple IBM Storage® Scale NSD file:

There are 7 nodes, each with 6 disk drives to be defined as NSDs. All information must be continuous with no extra spaces.

$ cp /var/lib/ambari-server/resources/gpfs_nsd.sample /var/lib/ambari-server/resources/gpfs_nsd
$ cat /var/lib/ambari-server/resources/gpfs_nsd

DISK|compute001.private.dns.zone:/dev/sdb,/dev/sdc,/dev/sdd,/dev/sde,/dev/sdf,/dev/sdg
DISK|compute002.private.dns.zone:/dev/sdb,/dev/sdc,/dev/sdd,/dev/sde,/dev/sdf,/dev/sdg
DISK|compute003.private.dns.zone:/dev/sdb,/dev/sdc,/dev/sdd,/dev/sde,/dev/sdf,/dev/sdg
DISK|compute005.private.dns.zone:/dev/sdb,/dev/sdc,/dev/sdd,/dev/sde,/dev/sdf,/dev/sdg
DISK|compute006.private.dns.zone:/dev/sdb,/dev/sdc,/dev/sdd,/dev/sde,/dev/sdf,/dev/sdg
DISK|compute007.private.dns.zone:/dev/sdb,/dev/sdc,/dev/sdd,/dev/sde,/dev/sdf,/dev/sdg

If you want to select disks such as SSD drives for metadata , add the label -meta to those disks.

In a simple NSD file, add the label meta for the disks that you want to use as metadata disks, as shown in the following example. If -meta is used, the Partition algorithm is ignored.

$ cat /var/lib/ambari-server/resources/gpfs_nsd

DISK|compute001.private.dns.zone:/dev/sdb-meta,/dev/sdc,/dev/sdd
DISK|compute002.private.dns.zone:/dev/sdb-meta,/dev/sdc,/dev/sdd
DISK|compute003.private.dns.zone:/dev/sdb-meta,/dev/sdc,/dev/sdd
DISK|compute005.private.dns.zone:/dev/sdb-meta,/dev/sdc,/dev/sdd
DISK|compute006.private.dns.zone:/dev/sdb,/dev/sdc,/dev/sdd
DISK|compute007.private.dns.zone:/dev/sdb,/dev/sdc,/dev/sdd

In the simple NSD file, /dev/sdb from compute001, compute002, compute003, and compute005 are specified as meta disks in the IBM Storage Scale file system.

The partition algorithm is ignored if the nodes listed in the simple NSD file do not match the set of nodes that will be used for the NodeManager service. If nodes that are not NodeManagers are in the NSD file or nodes that will be NodeManagers are not in the NSD file, no partitioning will be done.