0

I got some fiberchannel disk shelves including several 300Gb SCSI drives and a HP c7000 blade system with multiple blades in it.

As there is no storage controller (filer) available i have to do it without. The Idea is to use a blade that has access to the disks. (low cost solution)

I configured a centos on a server that has a qlogic fiberchannel port (2400). The disk shelves are directly connected to it. (in series with optical link)

When I boot the linux, and execute lsscsi the discs are recognised:

[0:0:0:0]    disk    NETAPP   X279_S15K5288F15 NA02  /dev/sda 
[0:0:1:0]    disk    NETAPP   X279_S15K5288F15 NA02  /dev/sdb 
[0:0:2:0]    disk    NETAPP   X279_S15K5288F15 NA02  /dev/sdc 
[0:0:3:0]    disk    NETAPP   X279_S15K5288F15 NA02  /dev/sdd 
[0:0:4:0]    disk    NETAPP   X279_S15K5288F15 NA02  /dev/sde 
[0:0:5:0]    disk    NETAPP   X279_S15K5288F15 NA02  /dev/sdf 
[0:0:6:0]    disk    NETAPP   X279_S15K5288F15 NA02  /dev/sdg 
[0:0:7:0]    disk    NETAPP   X279_S15K5288F15 NA02  /dev/sdh

...

The problem is that i cannot run any filesystem on it and cannot mount it (eg to use nfs to share the storage on the network.

How can I create a filesystem on the scsi discs so I can mount them on my linux? I tried fdisk /dev/sda it tells me no partision is available or no partition table?

Will it be possible to aggregate them into 1 giant partition and use some kind of RAID? (this is a second road to go, first a fs has to be created successfully)

Does someone know how to do this?

Falcon Momot
  • 24,975
  • 13
  • 61
  • 92
Jeffrey
  • 9
  • 1
  • 2
    Yes, it's completely possible... But it sounds like you have some pretty basic questions about filesystems and software RAID. Someone may answer, but you have a *lot* of options available to you. – ewwhite Jul 16 '14 at 22:08

1 Answers1

2

While I think that this is over your current knowledge and you would first need to understand basic concepts, such as what is a disk, a block device, a partition, a filesystem etc, here's the simplest solution.

Use pvcreate and create LVM on top of the disks:

pvcreate --pvmetadatacopies 3 /dev/sd{a,b,c,d,e,f,g,h}

Use vgcreate to create Volume Groups (you can probably create only one, depending on what you need to do):

vgcreate MyVolumeGroup /dev/sd{a,b,c,d,e,f,g,h}

Use lvcreate to create logical volumes, depending on your requirements. Here is an example creating a nice logical volume with 2 stripes and 2 mirrors for each written extent (somewhat similar to RAID 10):

lvcreate -i 2 -m 2 -n LOGICAL_VOLUME_NAME -L 10G MyVolumeGroup

Create a filesystem on the LVM:

mkfs.xfs /dev/mapper/MyVolumeGroup-LOGICAL_VOLUME_NAME

Mount the filesystem:

mount /dev/mapper/MyVolumeGroup-LOGICAL_VOLUME_NAME /mnt

IMPORTANT:

This is JUST an example that you can use so you can start testing your setup. You will need to learn a LOT about how LVM works in order to actually use such a setup in a production environment. You will need a LOT more knowledge that you currently have in order to efficiently use the type of storage that you are trying to use.

Florin Asăvoaie
  • 6,932
  • 22
  • 35