Skip to contentSkip to conversation

Insufficient free space: 1 extents needed, but only 0 available with 32.83g available

linux-lvm Brian J. Murrell 2 participants 3 messages Apr 1, 2013 – Apr 2, 2013

lore
Share & tools
lore
#0 lore raw

I have a VG:

VG #PV #LV #SN Attr VSize VFree
rootvol_tmp 1 26 0 wz--n- 74.40g 32.83g

Which has an LV in it which has a bad disk block on it. Per previous discussions here I have an LV into which I can remap PEs with bad blocks on them:

# lvdisplay --maps /dev/rootvol_tmp/badblocks
  --- Logical volume ---
  LV Name                /dev/rootvol_tmp/badblocks
  VG Name                rootvol_tmp
  LV UUID                oRCTFY-XNxh-3FA5-8lUG-FsOG-GjaO-giAcIu
  LV Write Access        read/write
  LV Status              available
  # open                 0
  LV Size                4.00 MiB
  Current LE             1
  Segments               1
  Allocation             inherit
  Read ahead sectors     auto
  - currently set to     256
  Block device           252:2

--- Segments ---
Logical extent 0 to 0:
Type linear
Physical volume /dev/sda2
Physical extents 2805 to 2805

As you can see, it has a PE in it already. I am trying to map another PE into it but getting an error:

# lvextend --alloc anywhere /dev/rootvol_tmp/badblocks -l +1 /dev/sda2:2812
  Extending logical volume badblocks to 8.00 MiB
  Insufficient free space: 1 extents needed, but only 0 available

Clearly there is something wrong with my syntax but I'm not quite sure what it is. Any ideas?

Much thanks!

b.

P.S. Please, no lectures about replacing the disk. I understand the risks of operating on a disk that is starting to show signs of wear. All of the data on this disk is backed up twice daily and anything lost between those backups will not be missed. It's just not feasible for me to replace this disk at this time. Thanks.

#1 lore raw
Quoted 33 lines · Brian J. Murrell

I have a VG:

VG #PV #LV #SN Attr VSize VFree
rootvol_tmp 1 26 0 wz--n- 74.40g 32.83g

Which has an LV in it which has a bad disk block on it. Per previous discussions here I have an LV into which I can remap PEs with bad blocks on them:

# lvdisplay --maps /dev/rootvol_tmp/badblocks
   --- Logical volume ---
   LV Name                /dev/rootvol_tmp/badblocks
   VG Name                rootvol_tmp
   LV UUID                oRCTFY-XNxh-3FA5-8lUG-FsOG-GjaO-giAcIu
   LV Write Access        read/write
   LV Status              available
   # open                 0
   LV Size                4.00 MiB
   Current LE             1
   Segments               1
   Allocation             inherit
   Read ahead sectors     auto
   - currently set to     256
   Block device           252:2

--- Segments ---
Logical extent 0 to 0:
Type linear
Physical volume /dev/sda2
Physical extents 2805 to 2805

As you can see, it has a PE in it already. I am trying to map another PE into it but getting an error:

Quoted 7 lines
# lvextend --alloc anywhere /dev/rootvol_tmp/badblocks -l +1 /dev/sda2:2812
   Extending logical volume badblocks to 8.00 MiB
   Insufficient free space: 1 extents needed, but only 0 available

Clearly there is something wrong with my syntax but I'm not quite sure what it is. Any ideas?

You said there is a LV containing the bad block. You can not have a block belonging to two LVs (for completeness sake: it is possible in snapshots but those shared extents can not be created like that ^)

*Is the extent 2812 free?*

*No*
Use something like pvmove --alloc anywhere /dev/sda2:2812 first.

Known issues: it does not work on mirrors and snapshots.
Workarounds? Split the mirror. Delete snapshot. Or you could try to hand-edit the metadata.

Also I am unsure what's the granularity of data it would skip - you may try to extract them first with e.g. ddrescue.

*Yes*
Then I have no idea except you may try specifying range /dev/sda2:2812-2812

Quoted 10 lines

Much thanks!

b.

P.S. Please, no lectures about replacing the disk. I understand the risks of operating on a disk that is starting to show signs of wear. All of the data on this disk is backed up twice daily and anything lost between those backups will not be missed. It's just not feasible for me to replace this disk at this time. Thanks.

I fully respect your choice, I myself have enough of throw away data like VMs...

Except you may want to consider following:

Just for convenience, if you can spare more than 2 blocks be more generous and disable larger extent of blocks - those disk errors like to spread... (or allocate to LV which loosing will be least )

Sometimes backups are not backups. To check if the backups work, have you ever tried to recover *all* data from those backups?

By size of the disk looks like notebook. Are you using disk encryption? (Loosing LUKS header could be rather painful! Have you backed up that as well? Otherwise you would need to restore everything - which may be impractical for example over slow network or with limited data.)

Similar applies to loosing LVM metadata.

And also depending on FS you may loose more than just few files.

-- Marian

Quoted 9 lines

_______________________________________________
linux-lvm mailing list
[email protected]
https://www.redhat.com/mailman/listinfo/linux-lvm
read the LVM HOW-TO at http://tldp.org/HOWTO/LVM-HOWTO/

#2 lore raw
Quoted 3 lines · Marian Csontos

You said there is a LV containing the bad block. You can not have a block belonging to two LVs

Ah ha! The missing link. Interestingly enough, as I was going through this process I had a nagging feeling that I somehow needed to release the block from the existing LV.

Quoted 1 line

Use something like pvmove --alloc anywhere /dev/sda2:2812 first.

Yeah. That did the trick.

Quoted 2 lines

I fully respect your choice, I myself have enough of throw away data like VMs...

Thanks.

Quoted 3 lines

Just for convenience, if you can spare more than 2 blocks be more generous and disable larger extent of blocks - those disk errors like to spread... (or allocate to LV which loosing will be least )

Yeah. I don't mind handling them piecemeal. Like I said, nothing on this machine is irreplaceable.

Quoted 2 lines

Sometimes backups are not backups. To check if the backups work, have you ever tried to recover *all* data from those backups?

But of course! You can't really call it a backup unless you have done a successful restore. I had to restore a machine from this same backup only a few weeks ago.

Quoted 1 line

By size of the disk looks like notebook.

Nope. Just old enough that that size disk was "common" when it was built.

Quoted 1 line

Are you using disk encryption?

Nope.

Quoted 1 line

(Loosing LUKS header could be rather painful!

Indeed. But not encrypting. In any case, losing the LUKS header might be just enough excuse to restore to bare-metal.

Quoted 1 line

And also depending on FS you may loose more than just few files.

Yeah. Pass on XFS and Reiserfs. Both have lost me too much data to use again. Fortunately ext{3,4} don't have such catastrophic failures just losing a block or two.

Cheers,
b.