replacing bad hard drive

[Heath Ramos]

I need to replace a bad hard drive and need to know the exact steps.

First of all....my set up. I am running hp-ux 11.0 on a D380. I received a notification email stating a hard drive is failing that included the hard drive path. I matched that path to the device and volume group (vg01 and /dev/dsk/c0t8d0). I also noted the logical volumes in the vg01 volume group. This isn't being mirrored.

What steps do I need to do in order to replace it (vgcfgrestore? vgchange? etc).

How do I know which drive in the drive bay is the hard drive in question?

[James R. Ferguson]

Hi Heath:

The best guide for dealing with disk failures is this:

http://www.docs.hp.com/en/5991-1236/When_Good_Disks_Go_Bad.pdf

Regards!

...JRF...

[Torsten. · Accepted solution]

As mentioned before, best advice is at
http://docs.hp.com/en/5991-1236/When_Good_Disks_Go_Bad.pdf

The disk is not mirroed, so shutdown, replace the disk, restore the config (vgcfgrestore); activate the vg, mount the volumes and restore the backup ...

[Heath Ramos]

thanks for the responses

how would I know which hard drive is the one failing, meaning the physical location in the server?

[James R. Ferguson]

Hi (again) Heath:

how would I know which hard drive is the one failing, meaning the physical location in the server?

If the drive is a local one and it has LED lights, you can try lighting the LED:

# dd if=/dev/rdsk/c0t8d0 of=/dev/null bs=1024k

...kill with Control_C...

Regards!

...JRF...

[Patrick Wallek]

In a D380, the disk with path t8 should be the 2nd drive from the top. From the top down, the drives are typically have SCSI ID's 5, 8, 9, 10, 11, I believe.

[Heath Ramos]

after reading that doc, I am going to do the following steps.

shutdown -h
replace 2nd hard drive from the top
boot the system
vgcfgrestore -n vg01 /dev/rdsk/c0t8d0
vgchange -a y vg01
***newfs [options] /dev/vg01/rlvol10
***newfs [options] /dev/vg01/rlvol13
mount /dev/vg01/lvol10 /db01
mount /dev/vg01/lvol13 /usr/oracle
restore from backup

I am not sure what options I need to put for the newfs commands

[James R. Ferguson]

Hi Heath:

Assuming that your filesystems are VxFS (JFS) ones, simply do:

# newfs -F vxfs -o largefiles /dev/vg01/rlvol10
# newfs -F vxfs -o largefiles /dev/vg01/rlvol13

The size of the filesystem will be that of the logical volume and the block size will be the default value (1024 bytes).

Setting the 'largefiles' bit enables files larger than 2GB to exist in the filesystem. Make sure that 'nolargefiles' is NOT set in '/etc/fstab' or you will not be able to mount the filesystems at boot.

Regards!

...JRF...

[Heath Ramos]

I was looking at my /etc/mnttab file and it says hfs for the logical volumes in question.

Should I just use that option?

newfs -F hfs /dev/vg01/rlvol10
newfs -F hfs /dev/vg01/rlvol13

[Heath Ramos]

do I want to be in single user mode when I boot back up after replacing the hard drive?

[James R. Ferguson]

Hi (again) Heath:

I'm surprised that you would still be running an HFS filesystem. VxFS ones are more robust and have been the default choice since 10.10 or 10.20.

That said, you can choose either. Substituting '-F hfs' for '-F vxfs' is correct. Edit your '/etc/fstab' appropriately. If you choose to create VxFS filesystems edit 'fstab' to look like:

/dev/vg01/lvol10 /db01 vxfs delaylog 0 2
/dev/vg01/lvol13 /usr/oracle vxfs delaylog 0 2

Regards!

...JRF...

[skt_skt]

Stanadard practice is only the /stand file system is of hfs type.Rest all is always can be vxfs type..

I would suggest to make it vxfs as you are getting a chance now.. I manage HP-UX servers(Includes 10.20.
;11.00 and 11.11 and 11.23) and i what i said true for all except for 11.23.

For 11.23 /stand is vxfs too.