re: rebuilding K570 server and "Model 20" storage array

[Gurcharan Sahota]

I'm having a lot of trouble reconfiguring a "Model 20" array. There is a vg00 group and there are 2 disks allocated to it

# strings /etc/lvmtab
/dev/vg00
/dev/dsk/c0t1d0
/dev/dsk/c1t0d0

i have setup vg01 and vg02 with the appropriate permissions and created a group file as well.

I have used pvcreate to initialize the other disks but receive the error
# pvcreate -f /dsk/rdsk/c0t1d2
Couldn't stat physical volume "/dsk/rdsk/c0t1d2":
Physical volume can be created only on whole disk.

I have had to unbind and then bind the disks on the array.

When i run a ioscan this is the output i get :
ioscan -fnkC disk
Class I H/W Path Driver S/W State H/W Type Description
======================================================================
disk 0 8/8.1.0 sdisk CLAIMED DEVICE DGC C2300WDR1
/dev/dsk/c0t1d0 /dev/rdsk/c0t1d0
disk 1 8/8.1.1 sdisk CLAIMED DEVICE DGC C2300WDR0
/dev/dsk/c0t1d1 /dev/rdsk/c0t1d1
disk 2 8/8.1.2 sdisk CLAIMED DEVICE DGC C2300WDR10
/dev/dsk/c0t1d2 /dev/rdsk/c0t1d2
disk 3 8/8.1.4 sdisk NO_HW DEVICE DGC C2300WDR5
/dev/dsk/c0t1d4 /dev/rdsk/c0t1d4
disk 4 8/8.1.5 sdisk NO_HW DEVICE DGC C2300WDR5
/dev/dsk/c0t1d5 /dev/rdsk/c0t1d5
disk 5 8/8.1.6 sdisk NO_HW DEVICE DGC C2300WDR1
/dev/dsk/c0t1d6 /dev/rdsk/c0t1d6
disk 6 8/8.1.7 sdisk NO_HW DEVICE DGC C2300WDR0
/dev/dsk/c0t1d7 /dev/rdsk/c0t1d7
disk 7 8/12.0.0 sdisk CLAIMED DEVICE DGC C2300WDR1
/dev/dsk/c1t0d0 /dev/rdsk/c1t0d0
disk 8 8/12.0.1 sdisk CLAIMED DEVICE DGC C2300WDR0
/dev/dsk/c1t0d1 /dev/rdsk/c1t0d1
disk 9 8/12.0.2 sdisk CLAIMED DEVICE DGC C2300WDR10
/dev/dsk/c1t0d2 /dev/rdsk/c1t0d2
disk 10 8/12.0.4 sdisk NO_HW DEVICE DGC C2300WDR5
/dev/dsk/c1t0d4 /dev/rdsk/c1t0d4
disk 11 8/12.0.5 sdisk NO_HW DEVICE DGC C2300WDR5
/dev/dsk/c1t0d5 /dev/rdsk/c1t0d5
disk 12 8/12.0.6 sdisk NO_HW DEVICE DGC C2300WDR1
/dev/dsk/c1t0d6 /dev/rdsk/c1t0d6
disk 13 8/12.0.7 sdisk NO_HW DEVICE DGC C2300WDR0
/dev/dsk/c1t0d7 /dev/rdsk/c1t0d7
disk 14 10/12/5.2.0 sdisk CLAIMED DEVICE PIONEER DVD-ROM DVD-30
3
/dev/dsk/c5t2d0 /dev/rdsk/c5t2d0

as you can see there are NO_HW's everywhere - how do i go about reclaiming these disks?

[Eugeny Brychkov]

1. if ioscan does not see LUNs (claims them as NO_HW) then these LUNs may be simply deleted. Check with grid manager if these LUNs are really there;
2. I notice there's LUN 3 is missing in sequence. Of course it's not a requirement, but as a test try creating LUN 3 and see if it will help. If it will help - then you probably are missing some SCSI cumulative patches...
Eugeny

[Gurcharan Sahota]

1 - where do i find grid manager?

2 - what is the command to recreate a lun?

thanks.

[Eugeny Brychkov]

To do all these actions you can use SAM.
Grid manager is a serial interface of disk array controllers. If you would connect to any controllers' serial port using 9600/8/N/1 (using, for example, 700/96 terminal) you'll get into controller firmware where you can view configuration and manage array.
I think for the first time try SAM. If it will not work - connect to serial port and check which LUNs are there. Please do not modify anything from grid manager
Eugeny

[Gurcharan Sahota]

I have tried binding the arrays thru sam, and found that some of the disks are showing 'CLAIMED'.

step 1 - thru sam

highlight disk array controller at 8/12.0, select 8 disks to bind using raid 1/0,

highlight disk array controller at 8/8.1,
select 8 disks to bind using raid 1/0,

why is it not claiming all the disks?

Class I H/W Path Driver S/W State H/W Type Description
======================================================================
disk 0 8/8.1.0 sdisk CLAIMED DEVICE DGC C2300WDR1
/dev/dsk/c0t1d0 /dev/rdsk/c0t1d0
disk 1 8/8.1.1 sdisk CLAIMED DEVICE DGC C2300WDR10
/dev/dsk/c0t1d1 /dev/rdsk/c0t1d1
disk 2 8/8.1.2 sdisk CLAIMED DEVICE DGC C2300WDR10
/dev/dsk/c0t1d2 /dev/rdsk/c0t1d2
disk 3 8/8.1.4 sdisk NO_HW DEVICE DGC C2300WDR5
/dev/dsk/c0t1d4 /dev/rdsk/c0t1d4
disk 4 8/8.1.5 sdisk NO_HW DEVICE DGC C2300WDR5
/dev/dsk/c0t1d5 /dev/rdsk/c0t1d5
disk 5 8/8.1.6 sdisk NO_HW DEVICE DGC C2300WDR1
/dev/dsk/c0t1d6 /dev/rdsk/c0t1d6
disk 6 8/8.1.7 sdisk NO_HW DEVICE DGC C2300WDR0
/dev/dsk/c0t1d7 /dev/rdsk/c0t1d7
disk 7 8/12.0.0 sdisk CLAIMED DEVICE DGC C2300WDR1
/dev/dsk/c1t0d0 /dev/rdsk/c1t0d0
disk 8 8/12.0.1 sdisk CLAIMED DEVICE DGC C2300WDR10
/dev/dsk/c1t0d1 /dev/rdsk/c1t0d1
disk 9 8/12.0.2 sdisk CLAIMED DEVICE DGC C2300WDR10
/dev/dsk/c1t0d2 /dev/rdsk/c1t0d2
disk 10 8/12.0.4 sdisk NO_HW DEVICE DGC C2300WDR5
/dev/dsk/c1t0d4 /dev/rdsk/c1t0d4
disk 11 8/12.0.5 sdisk NO_HW DEVICE DGC C2300WDR5
/dev/dsk/c1t0d5 /dev/rdsk/c1t0d5
disk 12 8/12.0.6 sdisk NO_HW DEVICE DGC C2300WDR1
/dev/dsk/c1t0d6 /dev/rdsk/c1t0d6
disk 13 8/12.0.7 sdisk NO_HW DEVICE DGC C2300WDR0
/dev/dsk/c1t0d7 /dev/rdsk/c1t0d7
disk 14 10/12/5.2.0 sdisk CLAIMED DEVICE PIONEER DVD-ROM DVD-30
3
/dev/dsk/c5t2d0 /dev/rdsk/c5t2d0

[Eugeny Brychkov]

What you see in the ioscan list are not physical disks, they are LUNs created within disk array. There're 3 LUNs claimed (0, 1 and 2) and 4 as no_hw (4, 5, 6, 7).
To create LUN you need to use 'bind' option and select unused disks. And give new LUN a number - I would like you to create LUN 3. And check if there're really LUNs 4, 5, 6, 7 exist in disk array...
Eugeny

[Gurcharan Sahota]

I have created the relevant luns, but when i go to run a vgcreate, i get the following error

# vgcreate /dev/vg01 /dev/dsk/c0t1d1
vgcreate: IO error on Physical Volume device /dev/dsk/c0t1d1

i pvcreated the disk and that runs ok
# pvcreate -f /dev/rdsk/c0t1d1
Physical volume "/dev/rdsk/c0t1d1" has been successfully created.

i then re-ran the vgcreate command and received the following message
# vgcreate /dev/vg01 /dev/dsk/c0t1d1
Increased the number of physical extents per physical volume to 4252.
vgcreate: Volume group "/dev/vg01" could not be created:
The path does not specify a valid physical volume.


the vgdisplay command returns the following -
# vgdisplay -v /dev/vg01
vgdisplay: Volume group "/dev/vg01" does not exist in the "/etc/lvmtab" file.
vgdisplay: Cannot display volume group "/dev/vg01".

any suggestions on where im going wrong?

[Gurcharan Sahota]

Thanks for your replies, but a colleague has pointed out (this is highlighted in the HPUX LVM Training Manual) that any low-level disk configuration work performed on a model 20 array must be done thru SAM. That probably explains why i have encountered all these problems! Starting over - hopefully the configuration work will run a lot smoother this time.