[linux-lvm] EMC Storage + LVM + XFS

[linux-lvm] EMC Storage + LVM + XFS

[Austin Gonyou]

I have a potential dilemma, and I'd like some feedback for those who might
be able to give it. If I have say 2-4 Fibre HBAs connected to an EMC
Symmetrix say 3930, and I create volumes, then disks are added some of
which who's LUN value is lower than disks already in the system, wouldn't
that cause all my sd values to change and move? I know if I add a scsi
disk into a LAS array that happens. So if this does happen, what's a good
procedure for recovering from it and making sure all my volumes are ok?

-- 
Austin Gonyou
Systems Architect, CCNA
Coremetrics, Inc.
Phone: 512-796-9023
email: austin@coremetrics.com

RE: [linux-lvm] EMC Storage + LVM + XFS

[S. Michael Denton]

 
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

vgscan is your friend :) (read my earlier post about moving an entire
bus around)

- -----Original Message-----
From: linux-lvm-admin@sistina.com
[mailto:linux-lvm-admin@sistina.com]On
Behalf Of Austin Gonyou
Sent: Thursday, 24 May 2001 22:46
To: linux-lvm@sistina.com
Subject: [linux-lvm] EMC Storage + LVM + XFS


I have a potential dilemma, and I'd like some feedback for those who
might
be able to give it. If I have say 2-4 Fibre HBAs connected to an EMC
Symmetrix say 3930, and I create volumes, then disks are added some
of
which who's LUN value is lower than disks already in the system,
wouldn't
that cause all my sd values to change and move? I know if I add a
scsi
disk into a LAS array that happens. So if this does happen, what's a
good
procedure for recovering from it and making sure all my volumes are
ok?

- -- 
Austin Gonyou
Systems Architect, CCNA
Coremetrics, Inc.
Phone: 512-796-9023
email: austin@coremetrics.com


_______________________________________________
linux-lvm mailing list
linux-lvm@sistina.com
http://lists.sistina.com/mailman/listinfo/linux-lvm
read the LVM HOW-TO at http://www.sistina.com/lvm/Pages/howto.html


-----BEGIN PGP SIGNATURE-----
Version: PGP 7.0.1

iQA/AwUBOw3KFXC+DNfF0nVpEQJoegCeK5/7drb1Nsy6En3yGMZm6NfYfq0An1Py
bgGoVMUyxZcDQDZypLNrHTIt
=M7RU
-----END PGP SIGNATURE-----

Re: [linux-lvm] EMC Storage + LVM + XFS

[Stephan Austermuehle]

On Thu, May 24, 2001 at 10:57:25PM -0400, S. Michael Denton wrote:

> vgscan is your friend :) (read my earlier post about moving an entire
> bus around)

This can only be a workaround because you cannot plug disks online
when you need to vgscan.

Stephan

Re: [linux-lvm] EMC Storage + LVM + XFS

[Austin Gonyou]

I don't think I'm clear on your statement. As far as I know, which is a
lot about EMC, you can definitely plug disks online. Having Linux see
them? Not necessarily so, but I believe the polling interval for the Fibre
driver will take care of that.

-- 
Austin Gonyou
Systems Architect, CCNA
Coremetrics, Inc.
Phone: 512-796-9023
email: austin@coremetrics.com

On Fri, 25 May 2001, Stephan Austermuehle wrote:

> On Thu, May 24, 2001 at 10:57:25PM -0400, S. Michael Denton wrote:
>
> > vgscan is your friend :) (read my earlier post about moving an entire
> > bus around)
>
> This can only be a workaround because you cannot plug disks online
> when you need to vgscan.
>
> Stephan
>
> _______________________________________________
> linux-lvm mailing list
> linux-lvm@sistina.com
> http://lists.sistina.com/mailman/listinfo/linux-lvm
> read the LVM HOW-TO at http://www.sistina.com/lvm/Pages/howto.html
>

Re: [linux-lvm] EMC Storage + LVM + XFS

[Stephan Austermuehle]

On Fri, May 25, 2001 at 09:47:34AM -0500, Austin Gonyou wrote:

> I don't think I'm clear on your statement. As far as I know, which is a
> lot about EMC, you can definitely plug disks online. Having Linux see
> them? Not necessarily so, but I believe the polling interval for the Fibre
> driver will take care of that.

Oh, of course the Symmetrix allows hot plugging of disks and online
mapping of hypers to one or more channels via a Bin upgrade or the
SDR. But at the moment I do not know any way to make Linux recognise
the additional disks without reloading (rebooting) the driver. But
even if there's a way to online detect new disks -- if a vgscan is
necessary to tell the Linux LVM about where its VGs have gone it isn't
a very smart solution because it often requires application down time.

Stephan

Re: [linux-lvm] EMC Storage + LVM + XFS

[Ragnar Kjørstad]

On Fri, May 25, 2001 at 05:09:08PM +0200, Stephan Austermuehle wrote:
> On Fri, May 25, 2001 at 09:47:34AM -0500, Austin Gonyou wrote:
> > I don't think I'm clear on your statement. As far as I know, which is a
> > lot about EMC, you can definitely plug disks online. Having Linux see
> > them? Not necessarily so, but I believe the polling interval for the Fibre
> > driver will take care of that.
> 
> Oh, of course the Symmetrix allows hot plugging of disks and online
> mapping of hypers to one or more channels via a Bin upgrade or the
> SDR. But at the moment I do not know any way to make Linux recognise
> the additional disks without reloading (rebooting) the driver. But
> even if there's a way to online detect new disks -- if a vgscan is
> necessary to tell the Linux LVM about where its VGs have gone it isn't
> a very smart solution because it often requires application down time.

I believe if you write things to /proc/scsi/scsi to make linux detect
new devices, they will be added _after_ the existing ones. The order
will not be messed up until next reboot, and then vgscan will take care
of it. Someone please correct me if I'm wrong :)


-- 
Ragnar Kj�rstad
Big Storage

Re: [linux-lvm] EMC Storage + LVM + XFS

[Austin Gonyou]

That is my initial assumption, but I only want to make sure that vgscan
_will_ in fact work they way it seems it should and that the private
region of all volumes or groups will be able to be read and recovered.

-- 
Austin Gonyou
Systems Architect, CCNA
Coremetrics, Inc.
Phone: 512-796-9023
email: austin@coremetrics.com

On Fri, 25 May 2001, Ragnar Kj�rstad wrote:

> On Fri, May 25, 2001 at 05:09:08PM +0200, Stephan Austermuehle wrote:
> > On Fri, May 25, 2001 at 09:47:34AM -0500, Austin Gonyou wrote:
> > > I don't think I'm clear on your statement. As far as I know, which is a
> > > lot about EMC, you can definitely plug disks online. Having Linux see
> > > them? Not necessarily so, but I believe the polling interval for the Fibre
> > > driver will take care of that.
> >
> > Oh, of course the Symmetrix allows hot plugging of disks and online
> > mapping of hypers to one or more channels via a Bin upgrade or the
> > SDR. But at the moment I do not know any way to make Linux recognise
> > the additional disks without reloading (rebooting) the driver. But
> > even if there's a way to online detect new disks -- if a vgscan is
> > necessary to tell the Linux LVM about where its VGs have gone it isn't
> > a very smart solution because it often requires application down time.
>
> I believe if you write things to /proc/scsi/scsi to make linux detect
> new devices, they will be added _after_ the existing ones. The order
> will not be messed up until next reboot, and then vgscan will take care
> of it. Someone please correct me if I'm wrong :)
>
>
>

Re: [linux-lvm] EMC Storage + LVM + XFS

[Heinz J. Mauelshagen]

On Fri, May 25, 2001 at 06:16:06PM -0500, Austin Gonyou wrote:
> That is my initial assumption, but I only want to make sure that vgscan
> _will_ in fact work they way it seems it should and that the private
> region of all volumes or groups will be able to be read and recovered.

Yes, it will (unless there's JAB ;-)

One of the major purposes of vgscan is to take care of changed io access
pathes to devices.

In case you add devices online using the scsi proc interface, you can use
them at their current access path and even though that likely will change after
a reboot (or reload of the driver) vgscan can catch up with the new
io configuration at startup before VG activation.

Regards,
Heinz    -- The LVM Guy --


> 
> -- 
> Austin Gonyou
> Systems Architect, CCNA
> Coremetrics, Inc.
> Phone: 512-796-9023
> email: austin@coremetrics.com
> 
> On Fri, 25 May 2001, Ragnar Kj�rstad wrote:
> 
> > On Fri, May 25, 2001 at 05:09:08PM +0200, Stephan Austermuehle wrote:
> > > On Fri, May 25, 2001 at 09:47:34AM -0500, Austin Gonyou wrote:
> > > > I don't think I'm clear on your statement. As far as I know, which is a
> > > > lot about EMC, you can definitely plug disks online. Having Linux see
> > > > them? Not necessarily so, but I believe the polling interval for the Fibre
> > > > driver will take care of that.
> > >
> > > Oh, of course the Symmetrix allows hot plugging of disks and online
> > > mapping of hypers to one or more channels via a Bin upgrade or the
> > > SDR. But at the moment I do not know any way to make Linux recognise
> > > the additional disks without reloading (rebooting) the driver. But
> > > even if there's a way to online detect new disks -- if a vgscan is
> > > necessary to tell the Linux LVM about where its VGs have gone it isn't
> > > a very smart solution because it often requires application down time.
> >
> > I believe if you write things to /proc/scsi/scsi to make linux detect
> > new devices, they will be added _after_ the existing ones. The order
> > will not be messed up until next reboot, and then vgscan will take care
> > of it. Someone please correct me if I'm wrong :)
> >
> >
> >
> 
> _______________________________________________
> linux-lvm mailing list
> linux-lvm@sistina.com
> http://lists.sistina.com/mailman/listinfo/linux-lvm
> read the LVM HOW-TO at http://www.sistina.com/lvm/Pages/howto.html

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

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] EMC Storage + LVM + XFS

[Austin Gonyou]

Right, and that's where I'm going. I can give app down time, but not just
because, oops, we for got this or that. Has to be methodically planned. As
long as all variables are known(as much as can be), then we can plan
properly.

-- 
Austin Gonyou
Systems Architect, CCNA
Coremetrics, Inc.
Phone: 512-796-9023
email: austin@coremetrics.com

On Fri, 25 May 2001, Stephan Austermuehle wrote:

> On Fri, May 25, 2001 at 09:47:34AM -0500, Austin Gonyou wrote:
>
> > I don't think I'm clear on your statement. As far as I know, which is a
> > lot about EMC, you can definitely plug disks online. Having Linux see
> > them? Not necessarily so, but I believe the polling interval for the Fibre
> > driver will take care of that.
>
> Oh, of course the Symmetrix allows hot plugging of disks and online
> mapping of hypers to one or more channels via a Bin upgrade or the
> SDR. But at the moment I do not know any way to make Linux recognise
> the additional disks without reloading (rebooting) the driver. But
> even if there's a way to online detect new disks -- if a vgscan is
> necessary to tell the Linux LVM about where its VGs have gone it isn't
> a very smart solution because it often requires application down time.
>
> Stephan
>
> _______________________________________________
> linux-lvm mailing list
> linux-lvm@sistina.com
> http://lists.sistina.com/mailman/listinfo/linux-lvm
> read the LVM HOW-TO at http://www.sistina.com/lvm/Pages/howto.html
>

RE: [linux-lvm] EMC Storage + LVM + XFS

[Christian, Chip]

You could play games with chains of:
	echo 'scsi-remove-single-device 1 0 0 2' >/proc/scsi/scsi
	echo 'scsi-add-single-device 1 0 0 1' >/proc/scsi/scsi
to get them to line up how you want.

You could use devfs and access your devices through
/dev/sd/...

On RedHat, at least, apparently you can use kudzu to do it (though I don't know how, not having explored this option).

-----Original Message-----
From: Austin Gonyou [mailto:austin@coremetrics.com]
Sent: Thursday, May 24, 2001 22:46
To: linux-lvm@sistina.com
Subject: [linux-lvm] EMC Storage + LVM + XFS


I have a potential dilemma, and I'd like some feedback for those who might
be able to give it. If I have say 2-4 Fibre HBAs connected to an EMC
Symmetrix say 3930, and I create volumes, then disks are added some of
which who's LUN value is lower than disks already in the system, wouldn't
that cause all my sd values to change and move? I know if I add a scsi
disk into a LAS array that happens. So if this does happen, what's a good
procedure for recovering from it and making sure all my volumes are ok?

-- 
Austin Gonyou
Systems Architect, CCNA
Coremetrics, Inc.
Phone: 512-796-9023
email: austin@coremetrics.com


_______________________________________________
linux-lvm mailing list
linux-lvm@sistina.com
http://lists.sistina.com/mailman/listinfo/linux-lvm
read the LVM HOW-TO at http://www.sistina.com/lvm/Pages/howto.html

RE: [linux-lvm] EMC Storage + LVM + XFS

[Austin Gonyou]

Yeah. I thought of all those things already, though thanks for the advice.
The DevFS idea is one I suppose I could try, but don't know if that's a
good idea. Thanks for your help!

-- 
Austin Gonyou
Systems Architect, CCNA
Coremetrics, Inc.
Phone: 512-796-9023
email: austin@coremetrics.com

On Fri, 25 May 2001, Christian, Chip wrote:

> You could play games with chains of:
> 	echo 'scsi-remove-single-device 1 0 0 2' >/proc/scsi/scsi
> 	echo 'scsi-add-single-device 1 0 0 1' >/proc/scsi/scsi
> to get them to line up how you want.
>
> You could use devfs and access your devices through
> /dev/sd/...
>
> On RedHat, at least, apparently you can use kudzu to do it (though I don't know how, not having explored this option).
>
> -----Original Message-----
> From: Austin Gonyou [mailto:austin@coremetrics.com]
> Sent: Thursday, May 24, 2001 22:46
> To: linux-lvm@sistina.com
> Subject: [linux-lvm] EMC Storage + LVM + XFS
>
>
> I have a potential dilemma, and I'd like some feedback for those who might
> be able to give it. If I have say 2-4 Fibre HBAs connected to an EMC
> Symmetrix say 3930, and I create volumes, then disks are added some of
> which who's LUN value is lower than disks already in the system, wouldn't
> that cause all my sd values to change and move? I know if I add a scsi
> disk into a LAS array that happens. So if this does happen, what's a good
> procedure for recovering from it and making sure all my volumes are ok?
>
>

RE: [linux-lvm] EMC Storage + LVM + XFS

[Christian, Chip]

To get the system to recognize a new disk without reloading the driver,
echo "scsi-add-single-device $controller $host $target $lun" > /proc/scsi/scsi

-----Original Message-----
From: Stephan Austermuehle [mailto:au@hcsd.de]
Sent: Friday, May 25, 2001 11:09
To: linux-lvm@sistina.com
Subject: Re: [linux-lvm] EMC Storage + LVM + XFS


On Fri, May 25, 2001 at 09:47:34AM -0500, Austin Gonyou wrote:

> I don't think I'm clear on your statement. As far as I know, which is a
> lot about EMC, you can definitely plug disks online. Having Linux see
> them? Not necessarily so, but I believe the polling interval for the Fibre
> driver will take care of that.

Oh, of course the Symmetrix allows hot plugging of disks and online
mapping of hypers to one or more channels via a Bin upgrade or the
SDR. But at the moment I do not know any way to make Linux recognise
the additional disks without reloading (rebooting) the driver. But
even if there's a way to online detect new disks -- if a vgscan is
necessary to tell the Linux LVM about where its VGs have gone it isn't
a very smart solution because it often requires application down time.

Stephan

_______________________________________________
linux-lvm mailing list
linux-lvm@sistina.com
http://lists.sistina.com/mailman/listinfo/linux-lvm
read the LVM HOW-TO at http://www.sistina.com/lvm/Pages/howto.html