[linux-lvm] Help, lvremove stuck in disk sleep for over 2 hours

[Tim Post]

Hello,

I have a lvremove process that has been stuck in state D (Disk sleep)
for over 2 hours, I am attaching a dump of its status at the end of this
e-mail.

Is there something that I can do to try and wake this up? A reboot on
this particular machine would be very, very bad. Processes in this state
can't be killed by signals. 

The LV that its trying to remove is named 6-e-7-sda1 and is in the
volume group named 'shelf1'.

If I list /dev/shelf1 , I see the LV as a broken symbolic link, also
verified that the LV is gone from /dev/mapper. I'm very concerned about
corrupting data on that VG so I'm hoping to get some advice on what to
do next.

LVM version:     2.02.26 (2007-06-15)
Library version: 1.02.20 (2007-06-15)
Driver version:  4.12.0
Using Ubuntu Hardy server, the VG is backed by an Areca ARC-1120 RAID 1

All other LVs in that group that are backing virtual machines seem to be
fine.

Is there any other information that I can provide?

Regards,
--Tim

Output below:

root at horizon5:/proc/10300# cat status
Name:   lvremove
State:  D (disk sleep)
Tgid:   10300
Pid:    10300
PPid:   1
TracerPid:      0
Uid:    0       0       0       0
Gid:    0       0       0       0
FDSize: 256
Groups: 0
VmPeak:    33704 kB
VmSize:    33704 kB
VmLck:     33704 kB
VmHWM:     15280 kB
VmRSS:     15280 kB
VmData:    10652 kB
VmStk:       272 kB
VmExe:       516 kB
VmLib:      2472 kB
VmPTE:        80 kB
Threads:        1
SigQ:   5/127286
SigPnd: 0000000000000100
ShdPnd: 0000000000024103
SigBlk: fffffffe7ffbfeff
SigIgn: 0000000000000000
SigCgt: 0000000180000000
CapInh: 0000000000000000
CapPrm: 00000000fffffeff
CapEff: 00000000fffffeff
voluntary_ctxt_switches:        120
nonvoluntary_ctxt_switches:     3