To main the Luster file system, We have to Install the Following 6 rpm package in Client,Object Storage Server(OSS),Meta Data Server(MDS),
To Install rpm package, use rpm -ivh
-i install
-v verose
-h hash
1)Even though we install the all package, e2fsprogs it is not install properly,
To check use rpm -qa command , So using rpm -Uvh update that package
2)Update the package
# rpm -Uvh e2fsprogs-1.41.10.sun2
Packages Are
1)e2fsprogs-1.41.10.sun2-0redhat.rhel5.x86_64.rpm
2)kernel-2.6.18-194.17.1.el5_lustre.1.8.5.x86_64.rpm
3)kernel-devel-2.6.18-194.17.1.el5_lustre.1.8.5.x86_64.rpm
4)lustre-1.8.5-2.6.18_194.17.1.el5_lustre.1.8.5.x86_64.rpm
5)lustre-ldiskfs-3.1.4-2.6.18_194.17.1.el5_lustre.1.8.5.x86_64.rpm
6)lustre-modules-1.8.5-2.6.18_194.17.1.el5_lustre.1.8.5.x86_64.rpm
Communcation Between Two Nodes,
For the communcation between the two node, We have to edit the in the modeprobe.conf file
1)we have to add entry in /etc/modprobe.conf
options lnet networks=tcp0(eth1)
#modprobe lnet [ update the driver]
List the Network
# lctl list_nids
IOC_LIBCFS_GET_NI error 100: Network is down
UP The Network
# lctl network up
LNET configured
# lctl list_nids
10.0.0.11@tcp0
Just like update the MDS, OSS and CLIENT,
To Check For Pinging Whether It is Pinging Or Not
[root@mds ~]# lctl ping 10.0.0.9@tcp
12345-0@lo
12345-10.0.0.9@tcp
These Are the Basic Luster Installation Step,
Note: In OSS and MDS we can create lustre file system using multiple
block device like /dev/sda1,/dev/sda2,/dev/sda3, but while mounting to the
client side we should mount with file system name. so it'll reflect as the
single partition from the multiple block device.
mount -t lustre [MDS node IP]:/[fsname] [mount Directory]
Lustre File system command
| #lctl | Lustre control Interface Command |
| List the Network | |
| Ping to the particular network. | |
| #lfs | Lustre File system Command. |
| ON quota,OFF quota,CHECK Quota,LIST Quota. [lfs quotaon MP,lfs quotaoff MP,lfs quota /home ,lfs quotacheck MP, ] LFS set striping and get striping.[lfs getstripe file. NOTE : MP = Mount Point | |
| List the file system [lfs df -h, | |
| #mkfs.lustre | Is used to create file system |
For Sample Luster Benchmark calculation
| 3.5 PB storage; 52 GB/s I/O throughput 131,072 processor cores |
| 1.73 PB storage; 40GB/s I/O throughput 62,976 processor cores |
| 340 TB Storage; 50GB/s I/O throughput 12,960 multi-core compute sockets |
| 10.5PB storage; 240 GB/s I/O throughput goal 265,708 processor cores |
0 Comments