[linux-lvm] LVM blocks discard for remove commands

[linux-lvm] LVM blocks discard for remove commands

[Yann Droneaud]

Hi,

Does lvremove, vgremove, pvremove issue discard (eg. TRIM) commands
in order to release LE/PE/metadata blocks from the underlying devices (SSD) ?

If LVM don't do it now, is it a planned feature ?

Meanwhile, which methods can be used to discard/TRIM blocks when removing
a LV, or a VG, or PV, without wiping the complete underlying device.

Something to feed hdparm --trim-sector-ranges once the LV, VG or PV is
removed for example ?

Regards

-- 
Yann Droneaud

Re: [linux-lvm] LVM blocks discard for remove commands

[Yann Droneaud]

Le mercredi 19 octobre 2011 à 12:45 +0200, Yann Droneaud a écrit :
> Hi,
> 
> Does lvremove, vgremove, pvremove issue discard (eg. TRIM) commands
> in order to release LE/PE/metadata blocks from the underlying devices (SSD) ?
> 

It seems the discard feature was added to LVM2 since version 2.02.85
(released on 29th April 2011), in WHATS_NEW:

  Add "devices/issue_discards" to lvm.conf.
  Issue discards on lvremove and lvreduce etc. if enabled and supported.

But the documentation is unclear about it. In example.conf.in /
lvm.conf.5.in:

"
Issue discards to a logical volumes's underlying physical volume(s) when
the logical volume is no longer using the physical volumes' space (e.g.
lvremove, lvreduce, etc).  Discards inform the storage that a region is
no longer in use.  Storage that supports discards advertise the protocol
specific way discards should be issued by the kernel (TRIM, UNMAP, or
WRITE SAME with UNMAP bit set).  Not all storage will support or benefit
from discards but SSDs and thinly provisioned LUNs generally do.  If set
to 1, discards will only be issued if both the storage and kernel
provide support.
"


Does it apply for the blocks holding meta data,
e.g. will pvremove issue discard requests for the PV meta data ?

Regards.

-- 
Yann Droneaud

Re: [linux-lvm] LVM blocks discard for remove commands

[Mike Snitzer]

On Wed, Oct 19 2011 at 10:09am -0400,
Yann Droneaud <yann@droneaud.fr> wrote:

> Le mercredi 19 octobre 2011 � 12:45 +0200, Yann Droneaud a �crit :
> > Hi,
> > 
> > Does lvremove, vgremove, pvremove issue discard (eg. TRIM) commands
> > in order to release LE/PE/metadata blocks from the underlying devices (SSD) ?
> > 
> 
> It seems the discard feature was added to LVM2 since version 2.02.85
> (released on 29th April 2011), in WHATS_NEW:
> 
>   Add "devices/issue_discards" to lvm.conf.
>   Issue discards on lvremove and lvreduce etc. if enabled and supported.
> 
> But the documentation is unclear about it. In example.conf.in /
> lvm.conf.5.in:
> 
> "
> Issue discards to a logical volumes's underlying physical volume(s) when
> the logical volume is no longer using the physical volumes' space (e.g.
> lvremove, lvreduce, etc).  Discards inform the storage that a region is
> no longer in use.  Storage that supports discards advertise the protocol
> specific way discards should be issued by the kernel (TRIM, UNMAP, or
> WRITE SAME with UNMAP bit set).  Not all storage will support or benefit
> from discards but SSDs and thinly provisioned LUNs generally do.  If set
> to 1, discards will only be issued if both the storage and kernel
> provide support.
> "
> 
> 
> Does it apply for the blocks holding meta data,
> e.g. will pvremove issue discard requests for the PV meta data ?

The discards are only issued for the physical extents that were used in
a volume's data area.  PV metadata isn't discarded, and really there
would be little benefit to doing so.

Re: [linux-lvm] LVM blocks discard for remove commands

[Stuart D. Gathman]

On Wed, 19 Oct 2011, Mike Snitzer wrote:

>> Does it apply for the blocks holding meta data,
>> e.g. will pvremove issue discard requests for the PV meta data ?
>
> The discards are only issued for the physical extents that were used in
> a volume's data area.  PV metadata isn't discarded, and really there
> would be little benefit to doing so.

The metadata is written over so that the PV is no longer recognized. 
I think that issuing a discard for the metadata would actually be
incorrect.

--
 	      Stuart D. Gathman <stuart@bmsi.com>
     Business Management Systems Inc.  Phone: 703 591-0911 Fax: 703 591-6154
"Confutatis maledictis, flammis acribus addictis" - background song for
a Microsoft sponsored "Where do you want to go from here?" commercial.