RE: Digest error handling

["Randy Jennings" <randyj@data-transit.com>]

Check out the section on Markers.  This question has been posted a few times
on the list in the last couple of weeks, starting with myself.  The authors
thought of a solution a long time ago, but it is easy to miss while reading.
If you don't use markers you would probably (with a very high probability)
have to close the connection.

Check the archives if you want more info.

Sincerely,
Randy Jennings
Data Transit
-----Original Message-----
From: owner-ips@ece.cmu.edu [mailto:owner-ips@ece.cmu.edu]On Behalf Of Amir
Grimberg
Sent: Thursday, April 18, 2002 7:45 AM
To: ips@ece.cmu.edu
Cc: Julian_Satran@il.ibm.com
Subject: Digest error handling


Hi All,
This is regarding digest error handling.

It says in draft 9 section 6.5 (iSCSI error handling and recovery -> digest
errors), that is the target receives a data PDU (solicited or unsolicited)
with header digest, it must silently discard the PDU. My question is how can
the target recover from this??
The problem is that there is no way to know the length of data payload after
the header. Should the target receive from the TCP socket 48 bytes of PDU
header in loop , until the CRC check is executed successfully ???
If yes, what happens if the data payload length is not divided by 48 bytes
(with no remainder) ? In this case, the CRC check will never be executed
successfully.

Thanks,
Amir Grimberg