Stale extents mirrored lvol with several disks

[R.O.]

Hi,

I have a lvol with a disk missing (it appears as "???" in lvdisplay. The lvol has 6 disks mirrored onto another 6 disks:

LV Name /dev/vg01/lvol1
VG Name /dev/vg01
LV Permission read/write
LV Status available/stale
Mirror copies 1


/dev/dsk/c4t0d0 4340 4340
/dev/dsk/c4t1d0 4340 4340
/dev/dsk/c4t2d0 4340 4340
/dev/dsk/c4t3d0 4340 4340
/dev/dsk/c4t8d0 17499 17499
/dev/dsk/c4t9d0 17499 17499
/dev/dsk/c5t0d0 4340 4340
/dev/dsk/c5t1d0 4340 4340
/dev/dsk/c5t2d0 4340 4340
.... MISSING DISK (c5t3d0)
/dev/dsk/c5t8d0 17499 17499
/dev/dsk/c5t9d0 17499 17499
....

30518 /dev/dsk/c5t1d0 04339 current /dev/dsk/c4t2d0 04339 current
30519 ??? 00000 stale /dev/dsk/c4t3d0 00000 current
30520 ??? 00001 stale /dev/dsk/c4t3d0 00001 current
30521 ??? 00002 stale /dev/dsk/c4t3d0 00002 current
...ETC

I have no access to the system right now. I know with two disks, I can do a "lvreduce -m 0 -A n –k /dev/vgname/lvname key", but in this case with more than one disk mirrored onto another disk, how can I do it? I mean, will lvdisplay -k show pvkeys 0-1-2-3 and so forth?

Regards,

[Torsten.]

No need to reduce anything, just follow the steps:

When_Good_Disks_Go_Bad_WP
http://bizsupport2.austin.hp.com/bc/docs/support/SupportManual/c01911837/c01911837.pdf

[R.O.]

Hi Torsten, nice to read you again!

The faulty disk is CLAIMED in ioscan, but has 0 bytes in diskinfo and vgdisplay and pvdisplay complain:

# pvdisplay -v /dev/dsk/c5t3d0
pvdisplay: Warning: couldn't query physical volume "/dev/dsk/c5t3d0":
The specified path does not correspond to physical volume attached to
this volume group


We have replaced the disk (now it has right size in diskinfo) and when we try to do a vgcfgrestore we get an error:

# vgcfgrestore -n /dev/vg01 /dev/rdsk/c5t3d0
vgcfgrestore: Warning: couldn't query physical volume "/dev/dsk/c5t3d0":
The specified path does not correspond to physical volume attached to
this volume group
vgcfgrestore: IO error on physical volume device "/dev/dsk/c5t3d0

Regards,

[Patrick Wallek]

What type of disk is it? Is it in some sort of Fibre enclosure (like an FC10 chassis)?

[Torsten.]

>> IO error on physical volume device "/dev/dsk/c5t3d0


If the disk is not the problem, then the problem is somewhere between the server and the disk, e.g. the cable.

Sorry, I don't know about your configuration details.

[R.O.]

Now I have this:

# vgcfgrestore -n /dev/vg01 /dev/rdsk/c5t3d0
vgcfgrestore: Warning: couldn't query physical volume "/dev/dsk/c5t3d0":
The specified path does not correspond to physical volume attached to
this volume group
Volume Group configuration has been restored to /dev/rdsk/c5t3d0
#

But vgdisplay still complains about disk:

vgdisplay: Warning: couldn't query physical volume "/dev/dsk/c5t3d0":
vgdisplay: Warning: couldn't query all of the physical volumes.

What could be happening? It is advisable to deactivate vg and activate again?

Regards,

[R.O.]

The disks are like this:

disk 12 0/12/0/0.3.0 sdisk CLAIMED DEVICE HP 18.2GST318452LC
/dev/dsk/c5t3d0 /dev/rdsk/c5t3d0

Regards,

[Bijeesh]

Hi,
Did you apply any commands like lvreduce,vgreduce before removing the faulty disk?

[R.O.]

Hi,

No, we simply replace the disk.

Regards,

[Patrick Wallek · Accepted solution]

Sometimes just activating the VG again will help.

# vgchange -a y /dev/vg01