Linux LVM pvmove -i trouble

linux-lvm Christian Hack 1 participant 2 messages Aug 4, 2004 1 message not in the archive

Share & tools

A message that is not in the archive (sent off-list or not yet received): [email protected]

Christian Hack

author

2004-08-04 02:14 UTC #1 lore raw

I emailed Heinz directly since I couldn't subscribe to the list. First hit in Google for "LVM mailing list" returns http://linux.msede.com/lvm_mlist/
which looks all fine and dandy, has up to date archives and instructions to subscribe which are obviously wrong. Perhaps the owner could point subscribers to the correct page below.

He replied directly also, so I'm moving this to the mailing list for posterity.

Quoted 66 lines

Quoted 1 line
I dug your email out of the archives since the linux-lvm
list seems totally
Quoted 3 lines
screwed up right now. I can't subscribe to the mailing list due to [email protected] not working properly and pretending to be [email protected].
You hit a very old URL ther.
Use http://www.redhat.com/mailman/listinfo/linux-lvm please.
Quoted 1 line
Anyway I have a dying drive with one PE that is bad on it.
I am trying to
Quoted 1 line
remove this last PE so I can remove the PV from the VG. I
used pvmove to get
Quoted 1 line
all but this bad PE off it with no troubles. "pvmove -i"
keeps bailing out
Quoted 4 lines
with this error:
[root@mythtv tools]# ./pvmove -i /dev/hdd pvmove -- moving physical extents in active volume group
"myth_volume_group"
Quoted 4 lines
pvmove -- WARNING: if you lose power during the move you may need
to restore your LVM metadata from backup!
pvmove -- do you want to continue? [y/n] y
/dev/myth_volume_group/group::/dev/myth_volume_group/myth_logi cal_volume:
Quoted 6 lines
1640 174981504, 0343 195035520
pvmove -- ERROR "Input/output error" copying extent from "/dev/hdd"
pvmove -- ERROR "Input/output error" moving physical extents
I have tried all combinations of pvmove, -i ,
--ignore_read_errors etc but
Quoted 1 line
it seems to simply ignore this option. I upgraded to 1.0.8
with the same
Quoted 3 lines
result. I see in the 1.0.8 notes it mentions:
"If you run 'pvmove' and run into device errors, you must
first run 'vgscan'
Quoted 1 line
before attempting 'pvmove -i'"
'-i' ignores read errors and tries to read as much data off the source device it can.
*But* it can't ignore a write error to the destination device of course.
Maybe that's your problem here ?
Quoted 2 lines
I did this with no success. I'm not even sure how vgscan
would achieve this
Quoted 1 line
anyway.
It scans the changed metadata in and creates /etc/lvmtab* in order to have the cached copy in sync with the ondosk copies for proper tool runs.
Quoted 7 lines
FYI I'm getting these sorts of errors in the logs
Aug  3 17:26:22 mythtv kernel: hdd: dma_intr: error=0x40 {
UncorrectableError }, LBAsect=175022778, high=10, low=7250618,
sector=175022721
Aug  3 17:26:22 mythtv kernel: end_request: I/O error, dev 
16:40 (hdd),
Quoted 2 lines
sector 175022721
Aug  3 17:26:25 mythtv kernel: hdd: dma_intr: status=0x51 { 

Quoted 8 lines

DriveReady
Quoted 3 lines
SeekComplete Error }
Aug  3 17:26:25 mythtv kernel: hdd: dma_intr: error=0x40 {
UncorrectableError }, LBAsect=175022778, high=10,
low=7250618, sector=1750
Hrm, no write io error on the destination device. Try running pvmove woth option v to see waht happens.

pvmove -i -vv /dev/hdd gives me this:

[root@mythtv mnt]# pvmove -i -vv /dev/hdd
pvmove -- checking name of source physical volume "/dev/hdd" pvmove -- locking logical volume manager
pvmove -- reading data of source physical volume from "/dev/hdd" pvmove -- checking volume group existence
pvmove -- reading data of volume group "myth_volume_group" from lvmtab pvmove -- checking volume group consistency of "myth_volume_group" pvmove -- searching for source physical volume "/dev/hdd" in volume group "myth_volume_group"
pvmove -- building list of possible destination physical volumes pvmove -- checking volume group activity
pvmove -- moving physical extents in active volume group "myth_volume_group" pvmove -- WARNING: if you lose power during the move you may need
to restore your LVM metadata from backup!
pvmove -- do you want to continue? [y/n] y
pvmove -- starting to move extents away from physical volume "/dev/hdd" pvmove -- checking for enough free physical extents in "myth_volume_group"
lv: /dev/myth_volume_group/myth_logical_volume[1] old_dev: 22:64 new_dev:
03:67 old_pe_sector: 174981504 new_pe_sector: 222953856
pvmove -- opening output physical volume "/dev/hdb3"
pvmove -- llseeking input physical volume "/dev/hdd"
pvmove -- llseeking output physical volume "/dev/hdb3"
pvmove -- /dev/hdd [PE 1334 [myth_logical_volume [LE 3121]] -> /dev/hdb3 [PE 1700] [1/1]
pvmove -- copying extent from /dev/hdd::/dev/myth_volume_group/myth_logical_volume(0x1640):174981504 to /dev/hdb3(0x0343):222953856 on disk /dev/myth_volume_group/group::/dev/myth_volume_group/myth_logical_volume:
1640 174981504, 0343 222953856
pvmove -- ERROR "Input/output error" copying extent from "/dev/hdd"

pvmove -- ERROR "Input/output error" moving physical extents

Quoted 8 lines

Quoted 1 line
How can I remove this PE. I don't care about the data
that's on it. I just
Quoted 1 line
want to get the failing drive out of the VG.
In case pvmove fails, only with surgery on the metadata -or- using the LVM2 tools which are capable to handle partital VGs and remove invalid PVs.
The later is the preferred way for obvious reasons.

Anyone else have suggestions before I go and get the LVM2 tools since pvmove -i should be working for me?

Christian Hack

author

2004-08-04 09:30 UTC #2 lore raw

Quoted 138 lines

I emailed Heinz directly since I couldn't subscribe to the list. First hit
in Google for "LVM mailing list" returns
http://linux.msede.com/lvm_mlist/
which looks all fine and dandy, has up to date archives and instructions to
subscribe which are obviously wrong. Perhaps the owner could point subscribers to the correct page below.
He replied directly also, so I'm moving this to the mailing list for posterity.
Quoted 19 lines
Quoted 1 line
I dug your email out of the archives since the linux-lvm
list seems totally
Quoted 3 lines
screwed up right now. I can't subscribe to the mailing list due to [email protected] not working properly and pretending to be [email protected].
You hit a very old URL ther.
Use http://www.redhat.com/mailman/listinfo/linux-lvm please.
Quoted 1 line
Anyway I have a dying drive with one PE that is bad on it.
I am trying to
Quoted 1 line
remove this last PE so I can remove the PV from the VG. I
used pvmove to get
Quoted 1 line
all but this bad PE off it with no troubles. "pvmove -i"
keeps bailing out
Quoted 4 lines
with this error:
[root@mythtv tools]# ./pvmove -i /dev/hdd pvmove -- moving physical extents in active volume group
Quoted 9 lines
"myth_volume_group"
Quoted 4 lines
pvmove -- WARNING: if you lose power during the move you may need
to restore your LVM metadata from backup!
pvmove -- do you want to continue? [y/n] y
/dev/myth_volume_group/group::/dev/myth_volume_group/myth_logi cal_volume:
Quoted 2 lines
1640 174981504, 0343 195035520
pvmove -- ERROR "Input/output error" copying extent from
"/dev/hdd"
Quoted 6 lines
Quoted 4 lines
pvmove -- ERROR "Input/output error" moving physical extents
I have tried all combinations of pvmove, -i ,
--ignore_read_errors etc but
Quoted 1 line
it seems to simply ignore this option. I upgraded to 1.0.8
Quoted 8 lines
with the same
Quoted 3 lines
result. I see in the 1.0.8 notes it mentions:
"If you run 'pvmove' and run into device errors, you must
first run 'vgscan'
Quoted 1 line
before attempting 'pvmove -i'"
'-i' ignores read errors and tries to read as much data off
the source
Quoted 11 lines
device it can.
*But* it can't ignore a write error to the destination device of course.
Maybe that's your problem here ?
Quoted 2 lines
I did this with no success. I'm not even sure how vgscan
would achieve this
Quoted 1 line
anyway.
It scans the changed metadata in and creates /etc/lvmtab*
in order to
Quoted 13 lines
have the cached copy in sync with the ondosk copies for proper tool runs.
Quoted 7 lines
FYI I'm getting these sorts of errors in the logs
Aug  3 17:26:22 mythtv kernel: hdd: dma_intr: error=0x40 {
UncorrectableError }, LBAsect=175022778, high=10, low=7250618,
sector=175022721
Aug  3 17:26:22 mythtv kernel: end_request: I/O error, dev 
16:40 (hdd),
Quoted 2 lines
sector 175022721
Aug  3 17:26:25 mythtv kernel: hdd: dma_intr: status=0x51 { 
Quoted 8 lines
DriveReady
Quoted 3 lines
SeekComplete Error }
Aug  3 17:26:25 mythtv kernel: hdd: dma_intr: error=0x40 {
UncorrectableError }, LBAsect=175022778, high=10,
low=7250618, sector=1750
Hrm, no write io error on the destination device. Try running pvmove woth option v to see waht happens.
pvmove -i -vv /dev/hdd gives me this:
[root@mythtv mnt]# pvmove -i -vv /dev/hdd
pvmove -- checking name of source physical volume "/dev/hdd" pvmove -- locking logical volume manager
pvmove -- reading data of source physical volume from "/dev/hdd" pvmove -- checking volume group existence
pvmove -- reading data of volume group "myth_volume_group" from lvmtab pvmove -- checking volume group consistency of "myth_volume_group" pvmove -- searching for source physical volume "/dev/hdd" in volume group
"myth_volume_group"
pvmove -- building list of possible destination physical volumes pvmove -- checking volume group activity
pvmove -- moving physical extents in active volume group "myth_volume_group"
pvmove -- WARNING: if you lose power during the move you may need
to restore your LVM metadata from backup!
pvmove -- do you want to continue? [y/n] y
pvmove -- starting to move extents away from physical volume "/dev/hdd"
pvmove -- checking for enough free physical extents in "myth_volume_group"
lv: /dev/myth_volume_group/myth_logical_volume[1] old_dev:
22:64 new_dev:
03:67 old_pe_sector: 174981504 new_pe_sector: 222953856 pvmove -- opening output physical volume "/dev/hdb3"
pvmove -- llseeking input physical volume "/dev/hdd"
pvmove -- llseeking output physical volume "/dev/hdb3" pvmove -- /dev/hdd [PE 1334 [myth_logical_volume [LE 3121]] -> /dev/hdb3 [PE
1700] [1/1]
pvmove -- copying extent from /dev/hdd::/dev/myth_volume_group/myth_logical_volume(0x1640):1 74981504 to
/dev/hdb3(0x0343):222953856 on disk /dev/myth_volume_group/group::/dev/myth_volume_group/myth_logi cal_volume:
1640 174981504, 0343 222953856
pvmove -- ERROR "Input/output error" copying extent from "/dev/hdd"
pvmove -- ERROR "Input/output error" moving physical extents
Quoted 4 lines
Quoted 1 line
How can I remove this PE. I don't care about the data
that's on it. I just
Quoted 1 line
want to get the failing drive out of the VG.

Quoted 9 lines

Quoted 4 lines
In case pvmove fails, only with surgery on the metadata -or- using the LVM2 tools which are capable to handle partital VGs and remove invalid PVs.
The later is the preferred way for obvious reasons.
Anyone else have suggestions before I go and get the LVM2 tools since pvmove
-i should be working for me?

Fixed my problem. It seems that my system was still using the old lvm runtime library that it was initially installed with (RH8 = LVM 1.0.3). Once I made sure the old libraries were replaced with with 1.0.8 libraries matching the binaries, pvmove -i successfully ignored the bad sectors (and told me so too) and I was able to remove the PV from the VG.