Re: iscsi : changes involving tgt portal group tag.

["Jim Hafner" <hafner@almaden.ibm.com>]

Santosh,

Some comments in line (the net is that I think you're right, but with a
slightly different reason in the first case).

Jim Hafner


Santosh Rao <santoshr@cup.hp.com>@t10.org on 03/01/2002 06:02:54 PM

Sent by:    owner-t10@t10.org


To:    IPS Reflector <ips@ece.cmu.edu>
cc:    T10 Reflector <t10@t10.org>
Subject:    iscsi : changes involving tgt portal group tag.



* From the T10 Reflector (t10@t10.org), posted by:
* Santosh Rao <santoshr@cup.hp.com>
*
All,

I believe iscsi needs to make 2 changes :

1) to mandate persistence of the portal group tag (and thus, persistence
of its scsi target port identifier).

2) to send the portal group tag (scsi target port identifier) as a part
of the iscsi login request.


The rationale for (1) is :
--------------------------
SAM-2 requires scsi port names to be persistent and world-wide-unique.
(SAM-2 Rev 22 Section 4.7.7). Quoting from SAM-2 :

"A scsi port name shall never change and may be used to persistently
identify a scsi initiator port or target port...".

iscsi defines scsi target port name as :

iscsi target name + NULL delimiter + 't' + NULL + (0 - 3) bytes of NULL
padding + 2 byte portal group tag (in big endian format).

In order for the above scsi target port name to meet SAM-2's requirement
of being persistent, the portal group tag needs to be persistent.

<JLH>
There are different ways that one can interpret this "persistent" rule. The
intent was that names shouldn't change over time when *identifying the same
object*.   What that means is that if the object changes (in any
discernable way), then the name should change.  So, the object can move to
a different piece of hardware, but it can still be named the same way.  If
the object changes, e.g., in this case, reconfigures as a different set of
network portals with different addressing elements, then I would think that
the name should change as well.   That's all the persistence one really
needs.

I'm not sure what that implies about your recommendation.  I don't see any
problem with it, either.
</JLH>

The rationale for (2) is :
--------------------------
Consider an initiator node establishing multiple sessions to a scsi tgt
port, with each session established to a subset of the network portals
within the tgt portal group.

Consider an iscsi transport event involving the re-configuration of
target portal groups on the iscsi target node. This may be preceeded by
the iscsi sessions seeing an async message "target requests logout",
followed by session logout, portal group re-configuration, and then, the
initiator re-establishes sessions.

Across a transport event that results in such session termination and
re-establishment, the initiator needs to authenticate that it is still
speaking to the same [i]scsi target port, in order to ensure that any
open/active I-T-L nexus traffic on that session is not incorrectly
routed to a wrong LUN after such a transport event.
<JLH>
Under these events, shouldn't all "open/active I_T_L traffic" have been
aborted, closed or otherwise ended?  So this shouldn't be an issue at all.

Note that the session end-points in such sessions  are individual
network portals within a portal group (tgt port) and a target
re-configuration involving the moving of network portals from 1 portal
group to another can occur.

Following such a re-configuration, if an initiator were to re-establish
a session to the same session end-points, since there is no addressing
information carried in the iscsi login request which carries the target
port identifier, the session may be established to the same network
portal, but under a different scsi target port and the I-T-L nexus
traffic incorrectly routed to a different scsi tgt port, resulting in
potential data corruption.

Other session oriented SCSI transports like FCP and SRP address this
problem by defining authentication schemes as well and/or explicitly
carrying the target port identifier in their login requests, which allow
them to identify such an authentication mis-match.

To prevent such authentication issues, iscsi can send the iscsi target
port identifier (portal group tag) explicitly in the login request, and
the login can be rejected by the target on a portal group tag mis-match.
(if the network portal does not belong to the addressed portal group
tag).
<JLH>
Did you mean for the initiator to send this TPGT?
</JLH>

Also, on any portal group re-configuration, iscsi targets MUST terminate
the session to avoid such scenarios as described above.
<JLH>
I would have thought that was implicit in the session logout.
</JLH>


Comments ?

<JLH>
Final comment:
I think you may have hit an odd corner case.  I think we need a rule that
says that (visibile) target port reconfiguration should involve new "names"
(either node or port). That such a reconfiguration should only occur after
all sessions have been dropped (with all the attendent clearing effects).
And that including the TPGT in the login with target reject if it isn't
correct should also be added.
</JLH>

Regards,
Santosh




--
##################################
Santosh Rao
Software Design Engineer,
HP-UX iSCSI Driver Team,
Hewlett Packard, Cupertino.
email : santoshr@cup.hp.com
Phone : 408-447-3751
##################################
*
* For T10 Reflector information, send a message with
* 'info t10' (no quotes) in the message body to majordomo@t10.org