failed disk

[navin]

Hi,
one of the hard disk went bad on an itanium server.now server is at no_hw state in ioscan.the server seems to have the lvm setup.
what is the procedure that needs to be followed up once the haulty disk is replaced.
The vgidisplay -v shows 4 lvms and the bg includes 2 disks.one of the disk went bad.How would i know the 2 disks involved are mirrored or concat.Also it is too late so we dont have vgcfgbackup.
thanks much

[R.K. # · Accepted solution]

Hello Navin,

Please confirm following:
1. Which VG (root or non-root)?
2. Disk is hot swappable or not?
3. lvdisplay -v /dev/vgname/lvolname | grep -i mirror #to check mirror

[R.K. #]

Hi Again,

else you can follow this:
http://docs.hp.com/en/5991-1236/When_Good_Disks_Go_Bad_WP.pdf

For any kind of disk replacement (for LVM disks).

[Johnson Punniyalingam]

always follow official documents from HP as mentioned below

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

[NouZ]

Hi
You confirm that whether there is Mirror or not
If it in mirrored before removing the disk you need to detach mirror of each Lv belongs to that disk using
#lvextend -m 0
then remove disk from VG
# Vgreduce
then you can replace the disk after that add to vg then extend lv to that specific PV

[Michael Steele_2]

Hi

Please post the contents of

ll /etc/lvmconf

I'm asking because vgcfgbackup will often work in the background after any LVM operation. You see the operation in syslog.log.

If you can't vgcfgrestore then

a) vgreduce bad disk out of vg
b) swap out bad disk for new
c) pvcreate
d) vgextend
e) lvextend -m 1 each lvol

You may have to vgreduce -f /dev/vg if a) above initially fails.