|
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] RE: iSCSI: Use of the A bitThanks for catching 2.5.3.4 conflict (it was not updated together with 9). It reads now: With the SNACK Request, the initiator requests retransmission of num-bered-responses or data from the target. A single SNACK request covers a contiguous set of missing items called a run of a given type of items (the type is indicate in a type field in the PDU header). The run is composed of an initial item (StatSN, DataSN, R2TSN) and the number of missed Status, Data, or R2T PDUs. For long data-in sequences, the target may request (at predefined minimum intervals) a positive acknowledgement for the data sent. A SNACK request with a type field that indicates ACK and the number of Data-In PDUs acknowl-edged conveys this positive acknowledgement. SNACKs can be rejected when the target does not have the data or status or has never sent it (error in error recovery :-)). Julo
9.4.3 I have a more general SNACK question. When would one use use the iSCSI, SCSI Response message with sense (reason -> SNACK rejected)? The reason for using this hasn't jumped out at me yet. Does it have to do with connection recovery? 9.16 States that SNACK support is optional. Since "ErrorRecoveryLevel" controls SNACK support, can something in the SNACK message description reference how this feature is negotiated? Also a cleanup is needed for SNACK- "RunLength".
Home Last updated: Fri Mar 15 15:18:14 2002 9137 messages in chronological order |