[linux-lvm] LVM Multi Path Support

[linux-lvm] LVM Multi Path Support

[Eduardo Dias]

[-- Attachment #1: Type: text/plain, Size: 214 bytes --]

Dear All,
 
I would like to know if LVM Linux has (or will) support to multi path links?
 
Regards,
 
Eduardo Dias



---------------------------------
Yahoo! Mail - O melhor e-mail do Brasil. Abra sua conta agora!

[-- Attachment #2: Type: text/html, Size: 464 bytes --]

Re: [linux-lvm] LVM Multi Path Support

[Joe Thornber]

On Wed, Mar 03, 2004 at 04:03:49PM -0300, Eduardo Dias wrote:
> Dear All,
>  
> I would like to know if LVM Linux has (or will) support to multi path links?

There is a v. old patch kicking around that adds some sort of multipath
support to LVM1.  Search this list to find it.  It hasn't ever been
part of an official release.

LVM2 doesn't support multipath but we're currently developing a
multipath target for device-mapper, and Christophe Varoqui is writing
a standalone userland tool to drive it.  The current thinking is that
the LVM2 tools will just use the devices that Christophes tool
creates.

- Joe

Re: [linux-lvm] LVM Multi Path Support

[Franc Carter]

Hi,

I have a related question.

I have just set up LVM on the LUNS from a HDS-9570 with a
qla-2300 driver in failover mode. This results in each
LUN being exposed to the operating system twice, once for
each controller. However when a controller fails the
card transparently maps the requests to the secondary
controller so that the original scsi target still works.

LVM appears to do the 'right thing for me' in that it recognises
that the the two PVs are one (I assume via the UUID) and ignore
the second LUN.

I'd like to know if this is just good luck, or if it's behaviour
I can rely ?

thanks

On Thursday 04 March 2004 20:30, Joe Thornber wrote:
> On Wed, Mar 03, 2004 at 04:03:49PM -0300, Eduardo Dias wrote:
> > Dear All,
> >  
> > I would like to know if LVM Linux has (or will) support to multi path 
links?
> 
> There is a v. old patch kicking around that adds some sort of multipath
> support to LVM1.  Search this list to find it.  It hasn't ever been
> part of an official release.
> 
> LVM2 doesn't support multipath but we're currently developing a
> multipath target for device-mapper, and Christophe Varoqui is writing
> a standalone userland tool to drive it.  The current thinking is that
> the LVM2 tools will just use the devices that Christophes tool
> creates.
> 
> - Joe
> 
> _______________________________________________
> linux-lvm mailing list
> linux-lvm@redhat.com
> https://www.redhat.com/mailman/listinfo/linux-lvm
> read the LVM HOW-TO at http://tldp.org/HOWTO/LVM-HOWTO/
> 
> 

-- 
Franc Carter     Ph:61-2-8374-5071      Fax: 61-2-8374-5070
Systems Manager, SIRCA Ltd              http://www.sirca.org.au/

DISCLAIMER: The contents of this email, inclusive of attachments, may
be legally privileged and confidential. Any unauthorised use of the
contents is expressly prohibited. If you have received this message in
error or are not the intended recipient, you should destroy the email
message along with any attachment(s).  Unintended recipients of this
email are prohibited from retaining, disclosing, distributing or using
any information contained herein.  This email is also subject to
copyright. No part of it should be reproduced, adapted or transmitted
without the written consent of the copyright owner.

Re: [linux-lvm] LVM Multi Path Support

[Joe Thornber]

On Fri, Mar 05, 2004 at 11:12:38AM +1100, Franc Carter wrote:
> LVM appears to do the 'right thing for me' in that it recognises
> that the the two PVs are one (I assume via the UUID) and ignore
> the second LUN.

Yes, LVM writes a label to the start of a PV to identify it.

> I'd like to know if this is just good luck, or if it's behaviour
> I can rely ?

I don't think it's defined which of the 2 devices LVM will choose,
does this matter ?  If so you could always set up a filter in lvm.conf
to remove one (probably advisable anyway).

- Joe

Re: [linux-lvm] LVM Multi Path Support

[Franc Carter]

Ahh, that makes sense (the undefined first PV) and explains
why the 'current-controller' is not the 'default-controller'
for one of the LUNs

It doesn't seem to matter if it uses the other device. I think it
just causes the array to do a tresspass and swap controllers. The
downside being that I can't reliably balance the I/O across the
two controllers. But that's not catastrophic.

I'm using lvm1, which I don't beleive has an lvm.conf - correct ?

thanks


On Fri, Mar 05, 2004 at 09:07:41AM +0000, Joe Thornber wrote:
> On Fri, Mar 05, 2004 at 11:12:38AM +1100, Franc Carter wrote:
> > LVM appears to do the 'right thing for me' in that it recognises
> > that the the two PVs are one (I assume via the UUID) and ignore
> > the second LUN.
> 
> Yes, LVM writes a label to the start of a PV to identify it.
> 
> > I'd like to know if this is just good luck, or if it's behaviour
> > I can rely ?
> 
> I don't think it's defined which of the 2 devices LVM will choose,
> does this matter ?  If so you could always set up a filter in lvm.conf
> to remove one (probably advisable anyway).
> 
> - Joe
> 
> _______________________________________________
> linux-lvm mailing list
> linux-lvm@redhat.com
> https://www.redhat.com/mailman/listinfo/linux-lvm
> read the LVM HOW-TO at http://tldp.org/HOWTO/LVM-HOWTO/

-- 
Franc Carter     Ph:61-2-8374-5071      Fax: 61-2-8374-5070
Systems Manager, SIRCA Ltd              http://www.sirca.org.au/

DISCLAIMER: The contents of this email, inclusive of attachments, may
be legally privileged and confidential. Any unauthorised use of the
contents is expressly prohibited. If you have received this message in
error or are not the intended recipient, you should destroy the email
message along with any attachment(s).  Unintended recipients of this
email are prohibited from retaining, disclosing, distributing or using
any information contained herein.  This email is also subject to
copyright. No part of it should be reproduced, adapted or transmitted
without the written consent of the copyright owner.

Re: [linux-lvm] LVM Multi Path Support

[Joe Thornber]

On Fri, Mar 05, 2004 at 09:27:55PM +1100, Franc Carter wrote:
> I'm using lvm1, which I don't beleive has an lvm.conf - correct ?

Correct.

- Joe

Re: [linux-lvm] LVM Multi Path Support

[Markus Baertschi]

I'd like to relate my experiences with multipath support and LVM.
We have an installation with several DB servers connected with
fiber to a SAN unit. All connections are doubled.
- Two fiber cards (Qlogic)
- Two switches
- SAN unit with dual controller (IBM FAStT 700)
We have tested the failover facilities provided by the qla2300-fo driver (pulled fibers) and they work. Each disk shows up once and the driver takes care of the failover facility.

However, we had one very nasty experience last summer. After rebooting (for an unrelated reason) the machines didn't came back up. The vgscan during boot dumped core. The reson was that somebody reconfigured the SAN unit and the controllers now had different WWNames. As a result the disks showed up twice and our (out of date) LVM could not handle it. Pulling one fiber/path temporarily fixed our problem. Upgrading to the latest LVM code fixed the coredump also. But we have not tested if failover would work that way.

Markus
-- 
  Markus Baertschi             Phone: ++41 (21) 807 1677
  Bas du Ross� 14b             Fax  : ++41 (21) 807 1678
  CH-1163, Etoy                Email: markus@markus.org
  Switzerland                  Homepage: www.markus.org