All of lore.kernel.org
 help / color / mirror / Atom feed
* FS directly on top of dm devices?
@ 2009-04-16 23:52 Scott Dungan
  2009-04-17  0:47 ` Christopher Chen
  2009-04-17 12:20 ` Romanowski, John (OFT)
  0 siblings, 2 replies; 5+ messages in thread
From: Scott Dungan @ 2009-04-16 23:52 UTC (permalink / raw)
  To: dm-devel

We recently had an engineer from a large storage vendor suggest that we 
make our file systems directly onto the multipathed devices without 
first creating underlying partitions or using LVM.  Having not run with 
this method in production before, we would like to gather as much 
information as possible first. Initial searches for documentation on 
this method has produced few results, short of a Suse/Novell document 
from 2005 under section 4:

http://support.novell.com/techcenter/sdb/en/2005/04/sles_multipathing.html

Does anyone have any experience with this or can point us to more 
documentation. thoughts or recommendations?

Thanks,

-- 

Scott A Dungan

^ permalink raw reply	[flat|nested] 5+ messages in thread

* Re: FS directly on top of dm devices?
  2009-04-16 23:52 FS directly on top of dm devices? Scott Dungan
@ 2009-04-17  0:47 ` Christopher Chen
  2009-04-17 15:55   ` Scott Dungan
  2009-04-17 12:20 ` Romanowski, John (OFT)
  1 sibling, 1 reply; 5+ messages in thread
From: Christopher Chen @ 2009-04-17  0:47 UTC (permalink / raw)
  To: device-mapper development

I run md over mpath dm'd devices, without partitioning. Of course, I
use dm-multipath to reach individual disks in a JBOD with two FC-AL
loops.

Lots of people like to do this to preserve stripe alignment.

I also use LVM, with the metadatasize option on creation of the PV, to
preserve stripe alignment.

Do you have a specific application in mind, like a database?

On Thu, Apr 16, 2009 at 4:52 PM, Scott Dungan <scott@gps.caltech.edu> wrote:
> We recently had an engineer from a large storage vendor suggest that we make
> our file systems directly onto the multipathed devices without first
> creating underlying partitions or using LVM.  Having not run with this
> method in production before, we would like to gather as much information as
> possible first. Initial searches for documentation on this method has
> produced few results, short of a Suse/Novell document from 2005 under
> section 4:
>
> http://support.novell.com/techcenter/sdb/en/2005/04/sles_multipathing.html
>
> Does anyone have any experience with this or can point us to more
> documentation. thoughts or recommendations?
>
> Thanks,
>
> --
>
> Scott A Dungan
>
>
> --
> dm-devel mailing list
> dm-devel@redhat.com
> https://www.redhat.com/mailman/listinfo/dm-devel
>



-- 
Chris Chen <muffaleta@gmail.com>
"I want the kind of six pack you can't drink."
-- Micah

^ permalink raw reply	[flat|nested] 5+ messages in thread

* RE: FS directly on top of dm devices?
  2009-04-16 23:52 FS directly on top of dm devices? Scott Dungan
  2009-04-17  0:47 ` Christopher Chen
@ 2009-04-17 12:20 ` Romanowski, John (OFT)
  2009-04-17 16:01   ` Scott Dungan
  1 sibling, 1 reply; 5+ messages in thread
From: Romanowski, John (OFT) @ 2009-04-17 12:20 UTC (permalink / raw)
  To: device-mapper development

Partitioning and LVM is optional, depends upon what you want to accomplish.
Sometimes I make the file system directly on the multipathed device, as in
mkfs /dev/mapper/mpatha

Other than having more paths to the LUN it's no different then making a file system on an unpartitioned scsi device as in
mkfs /dev/sda

> -----Original Message-----
> From: dm-devel-bounces@redhat.com [mailto:dm-devel-bounces@redhat.com]
> On Behalf Of Scott Dungan
> Sent: Thursday, April 16, 2009 7:52 PM
> To: dm-devel@redhat.com
> Subject: [dm-devel] FS directly on top of dm devices?
>
> We recently had an engineer from a large storage vendor suggest that we
> make our file systems directly onto the multipathed devices without
> first creating underlying partitions or using LVM.  Having not run with
> this method in production before, we would like to gather as much
> information as possible first. Initial searches for documentation on
> this method has produced few results, short of a Suse/Novell document
> from 2005 under section 4:
>
> http://support.novell.com/techcenter/sdb/en/2005/04/sles_multipathing.h
> tml
>
> Does anyone have any experience with this or can point us to more
> documentation. thoughts or recommendations?
>
> Thanks,
>
> --
>
> Scott A Dungan
>
>
> --
> dm-devel mailing list
> dm-devel@redhat.com
> https://www.redhat.com/mailman/listinfo/dm-devel


This e-mail, including any attachments, may be confidential, privileged or otherwise legally protected. It is intended only for the addressee. If you received this e-mail in error or from someone who was not authorized to send it to you, do not disseminate, copy or otherwise use this e-mail or its attachments.  Please notify the sender immediately by reply e-mail and delete the e-mail from your system.

^ permalink raw reply	[flat|nested] 5+ messages in thread

* Re: FS directly on top of dm devices?
  2009-04-17  0:47 ` Christopher Chen
@ 2009-04-17 15:55   ` Scott Dungan
  0 siblings, 0 replies; 5+ messages in thread
From: Scott Dungan @ 2009-04-17 15:55 UTC (permalink / raw)
  To: device-mapper development

Stripe alignment was the benefit we were thinking as well. These devices 
will be used for ext3 file systems that will store a mix of file sizes 
and types that are accessed by user-written and executed applications on 
an HPC cluster.

It is difficult to tune the file system or the underlying SAN LUNs for 
such a mix, so we are choosing our best estimate for general i/o.

It is comforting to know that others are using no partitioning.

Thanks.



Christopher Chen wrote:
> I run md over mpath dm'd devices, without partitioning. Of course, I
> use dm-multipath to reach individual disks in a JBOD with two FC-AL
> loops.
>
> Lots of people like to do this to preserve stripe alignment.
>
> I also use LVM, with the metadatasize option on creation of the PV, to
> preserve stripe alignment.
>
> Do you have a specific application in mind, like a database?
>
> On Thu, Apr 16, 2009 at 4:52 PM, Scott Dungan <scott@gps.caltech.edu> wrote:
>   
>> We recently had an engineer from a large storage vendor suggest that we make
>> our file systems directly onto the multipathed devices without first
>> creating underlying partitions or using LVM.  Having not run with this
>> method in production before, we would like to gather as much information as
>> possible first. Initial searches for documentation on this method has
>> produced few results, short of a Suse/Novell document from 2005 under
>> section 4:
>>
>> http://support.novell.com/techcenter/sdb/en/2005/04/sles_multipathing.html
>>
>> Does anyone have any experience with this or can point us to more
>> documentation. thoughts or recommendations?
>>
>> Thanks,
>>
>> --
>>
>> Scott A Dungan
>>
>>
>> --
>> dm-devel mailing list
>> dm-devel@redhat.com
>> https://www.redhat.com/mailman/listinfo/dm-devel
>>
>>     
>
>
>
>   

-- 

Scott A Dungan
Systems Administrator
GPS Division
California Institute of Technology
Phone: (626) 395-3170
Fax:   (626) 585-1917

^ permalink raw reply	[flat|nested] 5+ messages in thread

* Re: FS directly on top of dm devices?
  2009-04-17 12:20 ` Romanowski, John (OFT)
@ 2009-04-17 16:01   ` Scott Dungan
  0 siblings, 0 replies; 5+ messages in thread
From: Scott Dungan @ 2009-04-17 16:01 UTC (permalink / raw)
  To: device-mapper development

That is what we figured. Thanks for the reply.


Romanowski, John (OFT) wrote:
> Partitioning and LVM is optional, depends upon what you want to accomplish.
> Sometimes I make the file system directly on the multipathed device, as in
> mkfs /dev/mapper/mpatha
>
> Other than having more paths to the LUN it's no different then making a file system on an unpartitioned scsi device as in
> mkfs /dev/sda
>
>   
>> -----Original Message-----
>> From: dm-devel-bounces@redhat.com [mailto:dm-devel-bounces@redhat.com]
>> On Behalf Of Scott Dungan
>> Sent: Thursday, April 16, 2009 7:52 PM
>> To: dm-devel@redhat.com
>> Subject: [dm-devel] FS directly on top of dm devices?
>>
>> We recently had an engineer from a large storage vendor suggest that we
>> make our file systems directly onto the multipathed devices without
>> first creating underlying partitions or using LVM.  Having not run with
>> this method in production before, we would like to gather as much
>> information as possible first. Initial searches for documentation on
>> this method has produced few results, short of a Suse/Novell document
>> from 2005 under section 4:
>>
>> http://support.novell.com/techcenter/sdb/en/2005/04/sles_multipathing.h
>> tml
>>
>> Does anyone have any experience with this or can point us to more
>> documentation. thoughts or recommendations?
>>
>> Thanks,
>>
>> --
>>
>> Scott A Dungan
>>
>>
>> --
>> dm-devel mailing list
>> dm-devel@redhat.com
>> https://www.redhat.com/mailman/listinfo/dm-devel
>>     
>
>
> This e-mail, including any attachments, may be confidential, privileged or otherwise legally protected. It is intended only for the addressee. If you received this e-mail in error or from someone who was not authorized to send it to you, do not disseminate, copy or otherwise use this e-mail or its attachments.  Please notify the sender immediately by reply e-mail and delete the e-mail from your system.
>
>
> --
> dm-devel mailing list
> dm-devel@redhat.com
> https://www.redhat.com/mailman/listinfo/dm-devel
>   

-- 

Scott A Dungan
Systems Administrator
GPS Division
California Institute of Technology
Phone: (626) 395-3170
Fax:   (626) 585-1917

^ permalink raw reply	[flat|nested] 5+ messages in thread

end of thread, other threads:[~2009-04-17 16:01 UTC | newest]

Thread overview: 5+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2009-04-16 23:52 FS directly on top of dm devices? Scott Dungan
2009-04-17  0:47 ` Christopher Chen
2009-04-17 15:55   ` Scott Dungan
2009-04-17 12:20 ` Romanowski, John (OFT)
2009-04-17 16:01   ` Scott Dungan

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.