RE: iSCSI - Basic Question

["THALER,PAT (A-Roseville,ex1)" <pat_thaler@agilent.com>]

Joel,

I think Chhavi's question is when there is a header digest error, how does
the receiver know how much to discard and where in the stream to get the
header of the next PDU. 

Possible answers:

1) When a header digest error is received, stop receiving and close the
connection if FIM is not on. (Not a very desireable alternative.)

2) When a header digest error is received, assume the length may be okay
anyway and treat the information at that location as a header. If it passes
all header checks then continue processing for that connection. If it fails:

  2a) Stop receiving and close the connection if FIM is not on.
  2b) Start examining the stream for a set of bytes that passes the header
checks. (There is a possibility that one will get a false detection -
something in the data payload that happens to look like a header.)

3) If FIM is enabled, then the markers can be used to find a header and
continue processing.

I think that for a data stream without markers, 2a represents a reasonable
balance between the minimizing risk of false header detection and speeding
recovery from minor errors.

Pat

-----Original Message-----
From: Joel Evans [mailto:djjoelevans@hotmail.com]
Sent: Monday, April 15, 2002 8:20 AM
To: chhavi_garg@indiatimes.com; ips@ece.cmu.edu
Cc: Julian_Satran@il.ibm.com
Subject: Re: iSCSI - Basic Question



Chhavi,

You mentioned 'packet' length...  Are you referring a length issue with the 
packet, the segment, or the iSCSI payload?  The question is a little 
ambiguous - from the sound of it, it sounds as though you're asking about 
the length of the segment which would be handled by TCP error recovery - in 
which case, would be resent (no acknowledgement, etc).

Joel


>From: "chhavi_garg" <chhavi_garg@indiatimes.com>
>Reply-To: "chhavi_garg"<chhavi_garg@indiatimes.com>
>To: <ips@ece.cmu.edu>
>CC: <Julian_Satran@il.ibm.com>
>Subject: iSCSI - Basic Question
>Date: Mon, 15 Apr 2002 19:38:35 +0530
>MIME-Version: 1.0
>Received: from [128.2.136.200] by hotmail.com (3.2) with ESMTP id 
>MHotMailBE843884003F4004325D800288C8D6840; Mon, 15 Apr 2002 08:11:07 -0700
>Received: (from majordom@localhost)by ece.cmu.edu (8.11.0/8.10.2) id 
>g3FEALc10780for ips-outgoing; Mon, 15 Apr 2002 10:10:21 -0400 (EDT)
>Received: from WS0005.indiatimes.com ([203.199.93.15])by ece.cmu.edu 
>(8.11.0/8.10.2) with ESMTP id g3FEADw10751for <ips@ece.cmu.edu>; Mon, 15 
>Apr 2002 10:10:14 -0400 (EDT)
>Received: from 192.168.57.15 (a2 [192.168.57.22])by WS0005.indiatimes.com 
>(8.9.3/8.9.3) with SMTP id TAA07419;Mon, 15 Apr 2002 19:24:23 +0530
>From owner-ips@ece.cmu.edu Mon, 15 Apr 2002 08:11:40 -0700
>X-Authentication-Warning: ece.cmu.edu: majordom set sender to 
>owner-ips@ece.cmu.edu using -f
>Message-Id: <200204151354.TAA07419@WS0005.indiatimes.com>
>X-URL: http://indiatimes.com
>Sender: owner-ips@ece.cmu.edu
>Precedence: bulk
>
>
>
>
>Hi All,
>
>
>Section 6.5 of draft 11-95 says that,
>
>
>"When an initiator receives any iSCSI PDU with a header digest error, it 
>MUST discard the PDU."
>
>
>How will the initiator discard the packet as the packet length itself may 
>be wrong.
>
>
>How can the initiator recover from this condition?
>
>
>Should the initiator close this connection and start connection recovery?
>
>
>-regards,
>
>
>Chhavi.
>Get Your Private, Free E-mail from Indiatimes at  
>http://email.indiatimes.com
>Buy Music, Video, CD-ROM, Audio-Books and Music Accessories from 
>http://www.planetm.co.in





_________________________________________________________________
MSN Photos is the easiest way to share and print your photos: 
http://photos.msn.com/support/worldwide.aspx