How to replace faulty disk

[Rashid Ali]

I want to replace a faulty disk(c0t5d0) with a new one but the problem is this disk drive and root disk are in the same volume group (vg00). I run pvdisplay -v /dev/dsk/c0t5d0 and it shows /var is there. This faulty disk is still up and running and only when I run dd if=/dev/rdsk/c0t5d0 of=/dev/null then I got "Read error". Also, dmesg displays error message like
"SCSI: Read error -- dev: c 188 0x5000, errno: 124, resid: 262144,
blkno: 330752, sectno: 661504, offset: 338690048, bcount: 262144. "

I also checked that there is no free extends left on both disks. So I cannot run pvmove to move all files from faulty disk to another disk.

How can I replace the disk without reinstalling HP-UX 10.20?

Note: I am using HFS and don't have mirror.

Thanks in advance for any advices.

[Bill Hassell]

/var has critical information in it, specifically, all of the installed software. pvmove will fail just like dd if it encounters the bad areas on the disk. You can try moving the extents but this will might be a one time experiment. If pvmove fails, it will leave some of the extents on both disks. All this assumes you have enough space on your working disk to store the extents.

If you have an Ignite/UX backup, you could easily restore the system to the previous state.

If not, you'll have to choose between a cold reinstall or getting an external disk, extending it into your failing volume group and then trying the pvmove onto the new disk.

[Bruce Regittko_1]

Hi,

This may seem obvious but will your system let you add the new disk first, copy as much data as you can, and then remove the old disk?

You didn't mention what model 9000 system you have and some will only support 2 internal disks but if yours supports more... .

--Bruce