Open:

We will demo "How to deploy GPFS nodes massively using Diskless Remote Boot Linux".

Conetxt:

We start on 7 node and use 11 disks(6*160G + 5*320G=2.5T), then we dynamic add 3 disks(1*160G + 2*320G=0.8T). Final, we will see 3.3T in our environment.

• Installation

• DRBL Test

  $ sudo /opt/drbl/sbin/dcs
  

• GPFS Test

  $ sudo su
  $ cat gpfs.nodes
  gpfs00:quorum
  gpfs01:quorum
  gpfs02:
  gpfs03:
  gpfs04:
  gpfs05:
  gpfs06:
  gpfs07:
  
  $ mmcrcluster -N gpfs.nodes -p gpfs00 -s gpfs01 -r /usr/bin/ssh -R /usr/bin/scp
  $ mmlscluster
  $ mmlsnode -a
  
  $ cat gpfs.disks
  /dev/sda:gpfs01::dataAndMetadata::
  /dev/sdb:gpfs01::dataAndMetadata::
  /dev/sda:gpfs02::dataAndMetadata::
  /dev/sdb:gpfs02::dataAndMetadata::
  /dev/sda:gpfs03::dataAndMetadata::
  /dev/sdb:gpfs03::dataAndMetadata::
  /dev/sda:gpfs04::dataAndMetadata::
  /dev/sdb:gpfs04::dataAndMetadata::
  /dev/sda:gpfs05::dataAndMetadata::
  /dev/sdb:gpfs05::dataAndMetadata::
  /dev/sda:gpfs06::dataAndMetadata::
  
  $ mmcrnsd  -F gpfs.disks
  ($ mmcrnsd -F gpfs.disks -v no) --> if this disk ever used for gpfs
  $mmlsnsd
  
  $ mmstatup -a
  $ mmgetstate -a
  $ tsstatus
  
  $ mmcrfs /home/gpfs_mount gpfs0 -F gpfs.disks -B 1024K -r 1 -R 2 -m 1 -M 2 
  ($ mmcrfs /home/gpfs_mount gpfs0 -F gpfs.disks -B 1024K -r 2  -R 2 -m 2 -M 2 )  --> enable data replicate
  
  $ mmmount gpfs0 /home/gpfs_mount
  $ df -h
  
  $ cat gpfs.adddisks
  /dev/sdb:gpfs06::dataAndMetadata::
  /dev/sda:gpfs07::dataAndMetadata::
  /dev/sdb:gpfs07::dataAndMetadata::
  
  $ mmadddisk  gpfs0 -F gpfs.adddisks -r
  $ df -h