Creating a JBOD in my disk array

[Rick Christmas]

Hello,

  I have a rp3440 with 2 73gb internal drives, one is VG0 and contains my OS (HPUX 11.11i TCOE). The other has issues and I don't have time now to find out why.

  My FC card connects to a 2405 Disk Array that contains 15 146gb FC drives. I would like to configure the Disk Array to be a single JBOD of the 15x146GB. Everything I have tried to do gives me a different error message I don't understand.

  Can anyone tell me how to do this or outline the process?

  I tried adding one disk to a new volume group from the command line and got an error about 'extent', so I read in a forum to use SAM, SAM was cooking along pretty good and also errored out because of the extent, so I changed the extent, then it errored because of a 'unique identifier' that was the same as the other try that failed, even though I named them differently.

  Help would be greatly appreciated.

  Thank you in advance.

[Bill Hassell]

The DS2405 is a JBOD, that is Just A Bunch Of Disks. It is not array and has nothing inside to configure the disks in any manner except to allow access from either fibre connection. Now if this DS2405 was originally connected to a VA7400 (VA7410, etc), then the disk sectors have modified from 512 bytes to 520 bytes. So they cannot be used as simple JBOD disks at all. Here's a link to explain the 520 byte sector:

http://blog.sourcedirect.com/?s=disk+520

Now if you want to add a disk from the DS2405 to your vg00, it would have to have been created large enough to accomodate the extents for the largest disk (146 GB for the DS2405), but if you did not plan for larger disks, vg00 cannot handle the 146 GB disks. You would have to restore your system from an Ignite backup in order to add the disks and let Ignite resize vg00 extent counts.

[Rick Christmas]

Thank you for the reply. I have a vg00 that is my boot 76gb drive. A JBOD of 15x146 is my goal.

Rick

[Bill Hassell]

I believe what you are describing is a volume group with the 15 disks. Once created, all the disks may be setup as a single lvol and mounted. Use pvcreate on each of the device files for the DS2405 disks, then create the vg directory, the group file and use the vgcreate command. Once created, you can carve out an lvol using lvcreate and newfs, then mount the lvol as a single mount point of more than 2 TB.

[Rick Christmas]

I did pvcreate on each of the c2txd0 disks and got a affirmative response, but when I try to do the group file it starts the errors that the pvextent is too small

[Bill Hassell]

I did not give you detailed steps. pvcreate simply initializes the disks so they can be added to a volume group.

As I mentioned, you cannot use vgextend to add these 146 GB disks to your current vg00 because the disks are too big, that is, they require too many extents. When a volume group is first created, you can specify the maximum extent count, but if you don't, the count is the calculated based on the size of the current disks (the -e max_pe option). You can add many more 73 GB disks, but the maximum extent count (per disk)  won't allow adding the 146 GB disks.

So to create a new volume group, the steps are:

pvcreate (for each disk in the DS2405, which you have already done)
umask 077
mkdir /dev/vgbig   # vgbig is just the name I chose, you can use another name
ll /dev/*/group

## all the group files (perhaps just one) will be listed. You then pick an unused
##  minor number in the range 0x00..... to 0xff....
##  If there is just one group with 0x000000, then use 0x010000 as the new one

mknod /dev/vgbig/group c 64 0x010000
vgcreate vgbig /dev/dsk/c2t*d0

## Note: /dev/dsk/c2t*d0 should create a list of all the DS2405 disks

lvcreate -L 1000000 vgbig   # size is 1000 GB, just an arbitrary value
newfs -o largefiles /dev/vgbig/rlvol1
vgchange -a y vgbig
mkdir /bigdir             # arbitrary mountpoint name, choose your own
mount /dev/vgbig/lvol1 /bigdir
bdf                          # to see the new volume

 This will give you a new mountpoint to store files.

[Torsten.]

Just to add:

Usually you have redundant paths to each disk of this JBOD, so each disk as for example

/dev/dsk/c2t*d0
and
/dev/dsk/c4t*d0

So once you did a
# vgcreate ... /dev/dsk/c2t*d0
then do a

# vgextend ... /dev/dsk/c4t*d0

later.

You will see

/dev/dsk/c2t*d0
/dev/dsk/c4t*d0 Alternate Link

at the end of vgdisplay -v ...