multipath and READ_CAPACITY

multipath and READ_CAPACITY

[Hannes Reinecke]

Hi Christophe,

I've had a customer issue where someone complained that multipath 
would reject paths if they are in ALUA 'standby'.
As it turns out, SPC-3 does _not_ require the READ_CAPACITY command 
to be implemented if a target port is in standby.
Consecutively, linux would register this path with a capacity of 
'0', and multipath would reject the path in ev_add_path().

Which makes we wonder, do we actually _need_ the check for a matched 
capacity?
With VPD page 0x83 pretty much standard as of now (and I'm reasonbly 
sure we'll reject any device _not_ providing VPD page 0x83 anyway) 
we basically have to rely on that information.
IE if the WWID is identical we have assume the devices are 
identical. So it would make sense to just remove the check altogether.

Thoughts?

Cheers,

Hannes
-- 
Dr. Hannes Reinecke		      zSeries & Storage
hare@suse.de			      +49 911 74053 688
SUSE LINUX Products GmbH, Maxfeldstr. 5, 90409 Nürnberg
GF: J. Hawn, J. Guild, F. Imendörffer, HRB 16746 (AG Nürnberg)

Re: multipath and READ_CAPACITY

[Christophe Varoqui]

[-- Attachment #1.1: Type: text/plain, Size: 1433 bytes --]

This sanity check is useful because users can set arbitrary getuid scripts,
and thus corrupt data if their getuid callout is bogus.
Now, this getuid callout facility is a lot less publicized ...
I'd apply the patch, no problem on my side.


On Thu, Jul 24, 2014 at 10:46 AM, Hannes Reinecke <hare@suse.de> wrote:

> Hi Christophe,
>
> I've had a customer issue where someone complained that multipath would
> reject paths if they are in ALUA 'standby'.
> As it turns out, SPC-3 does _not_ require the READ_CAPACITY command to be
> implemented if a target port is in standby.
> Consecutively, linux would register this path with a capacity of '0', and
> multipath would reject the path in ev_add_path().
>
> Which makes we wonder, do we actually _need_ the check for a matched
> capacity?
> With VPD page 0x83 pretty much standard as of now (and I'm reasonbly sure
> we'll reject any device _not_ providing VPD page 0x83 anyway) we basically
> have to rely on that information.
> IE if the WWID is identical we have assume the devices are identical. So
> it would make sense to just remove the check altogether.
>
> Thoughts?
>
> Cheers,
>
> Hannes
> --
> Dr. Hannes Reinecke                   zSeries & Storage
> hare@suse.de                          +49 911 74053 688
> SUSE LINUX Products GmbH, Maxfeldstr. 5, 90409 Nürnberg
> GF: J. Hawn, J. Guild, F. Imendörffer, HRB 16746 (AG Nürnberg)
>

[-- Attachment #1.2: Type: text/html, Size: 2009 bytes --]

[-- Attachment #2: Type: text/plain, Size: 0 bytes --]