iSCSI: Target -> Initiator SNACK?

[Michael Schoberg <michael_schoberg@cnt.com>]

I'm having difficulty figuring out how a target handles CmdSN gaps it
detects (due to lost PDU, digest errors, etc.)  The draft reads like it is
not a Target issue and it's up to the initiator to recover from this.  How
are both the target and initiator are supposed to handle this situation?
Since the target cannot advance beyond it's expected next CmdSN (2.2.2.1),
how does the initiator detect the gap?  Through a timeout?  A target
initiated Nop-In?  

Thanks!

6.1.1  Usage of Retry

           By resending the same iSCSI command PDU ("retry") in the absence
of a 
           command acknowledgement or response, an initiator attempts to
"plug" 
           (what it thinks are) the discontinuities in CmdSN ordering on the
tar-
           get end.  Discarded command PDUs, due to digest errors, may have
cre-
           ated these discontinuities.
            
           Retry MUST NOT be used for reasons other than plugging command 
           sequence gaps.  In particular, all PDU retransmission (for data,
or 
           status) requests for a currently allegiant command in progress
must be 
           conveyed to the target using only the SNACK mechanism already 
           described.  This, however, does not constitute a requirement on
initi-
           ators to use SNACK.