Re: iSCSI: response to second login (with same ISID)

[Stephen Bailey <steph@cs.uchicago.edu>]

> Anybody else got any opinions?

I think it has to be:

> 2) CLOSE the pre-existing session (and abort all it's tasks) and then
> process the new login fresh.

This provides a mechanism for rapidly reclaiming session state in the
target.  For example, if an iSCSI adapter loses its mind (I'm sure we
all believe THAT will never happen), a login with the same SID will
reclaim the target state.

This capability is provided in FC, and certainly in the early days
(which was the first 4+ years) it was very important.  If you logged
in again, you knew the target would blast anything outstanding for the
initiator.  Granted, in FCP, you had a relatively short RA_TOV, but in
iSCSI RA_TOV could be pretty long.

That said, we can provide the capability in some other way, e.g. a
flag bit.  I don't care one way or the other, but if we don't provide
an explicit way to overload a session, then logging in with the same
SID must do it.

Steph