Re: working draft last comments before it goes out

["Matt Wakeley" <matt_wakeley@agilent.com>]

comments on the July 9  draft:

2.2.2. Ordering and iSCSI numbering:

"It is assumed that the SCSI layer frees all the resources
allocated for the task execution; the result summary, just enough
to rebuild the status PDU, will be kept by those iSCSI target
implementations that support status recovery after connection
failure."

This statement in 2.2.2 conflicts with 2.2.4 and 4.1
The target cannot "deallocate" all resources and only keep the result summary
around.  It must be able to send all the data if necessary (as it says in
4.1).



3.3. SCSI Response:

I'm not sure if it's required,  but the new "Res_len" & "Sense_len" fields are
not described.

3.10.4. Flags:

New text about numbering iscsi data messages appears here.  What's that got to
do with flags?  Also, I'm assuming that I->T transfers use CmdRN and I<-T
transfers use StatRN?

3.13 Login:

"RSA" is not defined.

Does there need to be some mention of IPSec?

There are so many "ways" to login, I think there might need to be a two step
login process.  The first step is more of a "query" that would indicate &
request what options are supported, and the second step would actually perform
the login.

It's lookin good!

Matt Wakeley