[linux-lvm] LV Fragmentation determination

[linux-lvm] LV Fragmentation determination

[Donald Thompson]

First off, can LV's become fragmented across a PV/VG? If yes then...

I'm looking for an easy way to determine how fragmented an LV might been
inside the VG. Is there any command I can run that'll give me nice
estimate on this?

Assuming theres no such command, can I determine this by looking at the
PE's displayed from an 'lvdisplay -v' command? I'm guessing it'd be pretty
easy to write a script that analyzes the PE numbers and looks for skips in
the sequence to give myself an idea of the amount of fragmention for the
LV. Or am I competely wrong thinking you can use PE's to make such a
determination?

I do shrinks and enlarges for different LV's sometimes on an almost daily
basis, and my VG's are getting to be very old. So I'm curious to see how
fragmentated my LVs have become.

TIA for any clues.

-Don

Re: [linux-lvm] LV Fragmentation determination

[Jon Bendtsen]

Donald Thompson wrote:
> 
> First off, can LV's become fragmented across a PV/VG? If yes then...

yes they can. But that depends on how you made it. The default is to
use the next free PE, but you can specify the PE's.

 
> I'm looking for an easy way to determine how fragmented an LV might been
> inside the VG. Is there any command I can run that'll give me nice
> estimate on this?

i dont know, but cant you use lvdisplay, pvdisplay, or the other
display commands to see the PE numbers ?

 
> Assuming theres no such command, can I determine this by looking at the
> PE's displayed from an 'lvdisplay -v' command? I'm guessing it'd be pretty
> easy to write a script that analyzes the PE numbers and looks for skips in
> the sequence to give myself an idea of the amount of fragmention for the
> LV. Or am I competely wrong thinking you can use PE's to make such a
> determination?

i think you can

 
> I do shrinks and enlarges for different LV's sometimes on an almost daily
> basis, and my VG's are getting to be very old. So I'm curious to see how
> fragmentated my LVs have become.

Properly much, but is that bad ? I mean, sure you have to move the head
every PE size, but how big are your PE size ? The default is 4MB.




JonB

Re: [linux-lvm] LV Fragmentation determination

[Heinz J . Mauelshagen]

On Wed, Nov 06, 2002 at 01:23:52AM -0700, Donald Thompson wrote:
> First off, can LV's become fragmented across a PV/VG? If yes then...

Yes, they can.

> 
> I'm looking for an easy way to determine how fragmented an LV might been
> inside the VG. Is there any command I can run that'll give me nice
> estimate on this?
> 
> Assuming theres no such command, can I determine this by looking at the
> PE's displayed from an 'lvdisplay -v' command? I'm guessing it'd be pretty
> easy to write a script that analyzes the PE numbers and looks for skips in
> the sequence to give myself an idea of the amount of fragmention for the
> LV. Or am I competely wrong thinking you can use PE's to make such a
> determination?
> 
> I do shrinks and enlarges for different LV's sometimes on an almost daily
> basis, and my VG's are getting to be very old. So I'm curious to see how
> fragmentated my LVs have become.

You can analyze the "lvdisplay -v" output on your LVs in question.
That shows you the mapping information and therefore, if your LVs are
fragemented.

> 
> TIA for any clues.
> 
> -Don
> 
> 
> _______________________________________________
> linux-lvm mailing list
> linux-lvm@sistina.com
> http://lists.sistina.com/mailman/listinfo/linux-lvm
> read the LVM HOW-TO at http://tldp.org/HOWTO/LVM-HOWTO/

-- 

Regards,
Heinz    -- The LVM Guy --

*** Software bugs are stupid.
    Nevertheless it needs not so stupid people to solve them ***

=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-

Heinz Mauelshagen                                 Sistina Software Inc.
Senior Consultant/Developer                       Am Sonnenhang 11
                                                  56242 Marienrachdorf
                                                  Germany
Mauelshagen@Sistina.com                           +49 2626 141200
                                                       FAX 924446
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-

Re: [linux-lvm] LV Fragmentation determination

[Anders Widman]

>> First off, can LV's become fragmented across a PV/VG? If yes then...

> Yes, they can.

>> 
>> I'm looking for an easy way to determine how fragmented an LV might been
>> inside the VG. Is there any command I can run that'll give me nice
>> estimate on this?

> You can analyze the "lvdisplay -v" output on your LVs in question.
> That shows you the mapping information and therefore, if your LVs are
> fragemented.

Will LVM include a defragment procedure?

Re: [linux-lvm] LV Fragmentation determination

[Joe Thornber]

On Wed, Nov 06, 2002 at 02:17:00PM +0100, Anders Widman wrote:
> >> First off, can LV's become fragmented across a PV/VG? If yes then...
> 
> > Yes, they can.
> 
> >> 
> >> I'm looking for an easy way to determine how fragmented an LV might been
> >> inside the VG. Is there any command I can run that'll give me nice
> >> estimate on this?
> 
> > You can analyze the "lvdisplay -v" output on your LVs in question.
> > That shows you the mapping information and therefore, if your LVs are
> > fragemented.
> 
> Will LVM include a defragment procedure?

LVM2 will at some point, in that arbitrary remappings will be
supported.  However I'm not convinced that this is a big problem, the
extents are 4M after all.

- Joe

Re: [linux-lvm] LV Fragmentation determination

[Heinz J . Mauelshagen]

On Wed, Nov 06, 2002 at 01:38:03PM +0000, Joe Thornber wrote:
> On Wed, Nov 06, 2002 at 02:17:00PM +0100, Anders Widman wrote:
> > >> First off, can LV's become fragmented across a PV/VG? If yes then...
> > 
> > > Yes, they can.
> > 
> > >> 
> > >> I'm looking for an easy way to determine how fragmented an LV might been
> > >> inside the VG. Is there any command I can run that'll give me nice
> > >> estimate on this?
> > 
> > > You can analyze the "lvdisplay -v" output on your LVs in question.
> > > That shows you the mapping information and therefore, if your LVs are
> > > fragemented.
> > 
> > Will LVM include a defragment procedure?
> 
> LVM2 will at some point, in that arbitrary remappings will be
> supported.  However I'm not convinced that this is a big problem, the
> extents are 4M after all.

True.

BTW: the default got changed to 32MB to support up to 2TB LVs in
     LVM 1.0.6 by default rather than by "vgchange -s" which makes the
     impact even smaller.

> 
> - Joe
> 
> _______________________________________________
> linux-lvm mailing list
> linux-lvm@sistina.com
> http://lists.sistina.com/mailman/listinfo/linux-lvm
> read the LVM HOW-TO at http://tldp.org/HOWTO/LVM-HOWTO/

-- 

Regards,
Heinz    -- The LVM Guy --

*** Software bugs are stupid.
    Nevertheless it needs not so stupid people to solve them ***

=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-

Heinz Mauelshagen                                 Sistina Software Inc.
Senior Consultant/Developer                       Am Sonnenhang 11
                                                  56242 Marienrachdorf
                                                  Germany
Mauelshagen@Sistina.com                           +49 2626 141200
                                                       FAX 924446
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-