RE: iscsi: CRN support not required. [was :Re: [Fwd: iSCSI: items discussed at WG meeting]]

["Julian Satran" <Julian_Satran@il.ibm.com>]

David,

I am afraid that the current text of SAM2 will not allow us to remove CRN. It states explicitly that "when the optional CRN argument is used all sequential commands of an I_T_L nexus shall include a CRN argument ..." (pg 70) (only the 0 value is reserved to be defined by the transport).

I agree that the use is defined in a very fuzzy manner and how the option is activated should be better handled by T10
but I don't think that we can remove it (unless T10 changes it's status).

You may recall that this requirement was also the base we introduced it.

As it stands - in iSCSI - CRN is "supported" when the ULP has a way to convey it to iSCSI.

We should make clear to T10 that if the option is to be controlled by a mode page that has to be a SCSI not a transport mode page.

Regards,
Julo

Black_David@emc.com Sent by: owner-ips@ece.cmu.edu 04-04-02 04:57 Please respond to Black_David

To:        John Hufferd/San Jose/IBM@IBMUS, ips@ece.cmu.edu         cc:                 Subject:        RE: iscsi: CRN support not required. [was :Re: [Fwd: iSCSI: items         discussed at WG meeting]]

> Every thing I have seen here, has nothing to do with an issue that is
> broken in iSCSI.  I suggest that whomever wants to take it up with T10 do
> so, and lets move on.  This is not a issue to be endlessly debated in this
> list.

I agree with John that this is not broken.  I seem to recall that CRN
was added as something that SAM-2 had invented and hence iSCSI had to carry.
Given the inconsistent state of its support and no serious request of
the form "have to have CRN or else something really important breaks",
I think the right thing to do is take CRN out, and reserve the field that
carried it in case we have to put it back, especially if the argument for
CRN involves T10 changing its definition (we should let T10 figure this
out first).  LUN-specific mode pages for iSCSI are a non-starter, and
if CRN requires them, iSCSI is unlikely to ever support CRN.  Unless there's
a pointed objection to this, Julian should take this message as an
instruction
to remove CRN from the iSCSI draft.

The issues regarding CRN usage/behavior across FC/iSCSI need to get written
up in the bridge/gateway draft that's being prepared.

Thanks,
--David
---------------------------------------------------
David L. Black, Senior Technologist
EMC Corporation, 42 South St., Hopkinton, MA  01748
+1 (508) 249-6449 *NEW*      FAX: +1 (508) 497-8500
black_david@emc.com         Cell: +1 (978) 394-7754
---------------------------------------------------