1.Installing the GPFS software
1.1 Installing the GPFS by source code
Need use the vnc or xmanager to run this commond#tar -xvf GPFS_3.5_WIN-LIN.tar#./gpfs_install-3.5.0-0_x86_ Install depend RPM
yum install compat-libstdc++-33.x86_ kernel-headers kernel-devel gcc make gcc-c++ rpm-buildInstall and upgrade the gpfs rpm
#rpm -ivh /usr/lpp/mmfs/3.5/gpfs*.rpm
#tar -zxvf GPFS-3.5.0.11-x86_-Linux.tar.gz #rpm -Uvh gpfs*.rpm
Compile the kernel layer gpfs files and packaged into rpm format#cd /usr/lpp/mmfs/src#make Autoconfig#make World#make InstallImages #make rpm
Install the compiled rpm packages
#rpm -ivh /root/rpmbuild/RPMS/x86_/gpfs.gplbin-2.6.32-279.el6.x86_-3.5.0-11.x86_.rpm
Changing environment parametervi /etc/profile
# Path manipulation
if [ \"$EUID\" = \"0\" ]; then pathmunge /sbin pathmunge /usr/sbin
pathmunge /usr/local/sbin
pathmunge /usr/lpp/mmfs/bin #pls add the linefi
1.2 Install the GPFS by RPM
Need use the vnc or xmanager to run this commond#tar -xvf GPFS_3.5_WIN-LIN.tar#./gpfs_install-3.5.0-0_x86_
Install depend RPM
yum install compat-libstdc++-33.x86_ kshInstall and upgrade the gpfs rpm
#rpm -ivh /usr/lpp/mmfs/3.5/gpfs*.rpm
#tar -zxvf GPFS-3.5.0.11-x86_-Linux.tar.gz #rpm -Uvh gpfs*.rpm
#rpm -ivh gpfs.gplbin-2.6.32-279.el6.x86_-3.5.0-11.x86_.rpm
Changing environment parametervi /etc/profile
# Path manipulation
if [ \"$EUID\" = \"0\" ]; then pathmunge /sbin pathmunge /usr/sbin
pathmunge /usr/local/sbin
pathmunge /usr/lpp/mmfs/bin #pls add the linefi
2.Creating your GPFS cluster
#vi /tmp/gpfsprofilenova01:quorum-managernova02:quorum-managernova03:quorum-managernova04:quorum-managernova05:quorum-manager
#mmcrcluster -N /tmp/gpfsprofile -p gpfs01 -s gpfs02 -r /usr/bin/ssh -R /usr/bin/scp -C bjcugpfs
3.Create nsd disk
3.1 Create the logical volume (on all node)
[root@nova03 ~]# vgdisplay --- Volume group --- VG Name vg_nova03 System ID Format lvm2 Metadata Areas 1 Metadata Sequence No 3 VG Access read/write VG Status resizable MAX LV 0 Cur LV 2 Open LV 2 Max PV 0 Cur PV 1 Act PV 1
VG Size 834.71 GiB PE Size 4.00 MiB Total PE 213685
Alloc PE / Size 13824 / 54.00 GiB Free PE / Size 199861 / 780.71 GiB
VG UUID Wc4GwZ-sEgm-lI7Z-XdyV-LN1k-eby4-erYzo7
#lvcreate -L 750G -n dataonly vg_nova03#lvcreate -L 30G -n metedata vg_nova03
3.2LV formatting (on all node)
#mkfs -t ext3 /dev/vg_nova03/dataonly#mkfs -t ext3 /dev/vg_nova03/metedata# cat /proc/partitions
3.3 Create nsd disk (on manager node)
#vi /tmp/nsd.conf%nsd: device=dm-2 nsd=disk01 servers=nova01 usage=dataOnly failuregroup=1 pool=datapool%nsd: device=dm-3 nsd=disk02 servers=nova01 usage=metadataOnly failuregroup=1pool=system%nsd: device=dm-2 nsd=disk03 servers=nova02 usage=dataOnly failuregroup=2pool=datapool%nsd: device=dm-3 nsd=disk04 servers=nova02 usage=metadataOnly failuregroup=2pool=system%nsd: device=dm-2 nsd=disk05servers=nova03 usage=dataOnly failuregroup=3 pool=datapool%nsd: device=dm-3 nsd=disk06
servers=nova03 usage=metadataOnly failuregroup=3pool=system%nsd: device=dm-2 nsd=disk07 servers=nova04 usage=dataOnly failuregroup=4pool=datapool%nsd: device=dm-3 nsd=disk08 servers=nova04 usage=metadataOnly failuregroup=4pool=system%nsd: device=dm-2 nsd=disk09 servers=nova05 usage=dataOnly failuregroup=5pool=datapool%nsd: device=dm-3 nsd=disk10 servers=nova05 usage=metadataOnly failuregroup=5pool=system
#mmcrnsd -F /tmp/nsd.conf
4.Create filesystem
4.1Start up the GPFS cluster
#mmstartup -a
4.2 create the filesystem
mmcrfs gpfs -F /tmp/nsd.conf -A yes -B 512K -j cluster -m 2 -r 2 -M 3 -R 3 --block-group-factor 128 -s failuregroupRoundRobin --write-affinity-depth 1 -T /var/nova --metadata-block-size 256K Explanation：
A write affinity depth of 1 indicates that the first copy is written to the writer node. For the
remaining replicas, each replica is striped across the disks in a cyclical fashion with the restriction that no two disks are in the same failure group.
