SORT BY:

LIST ORDER
THREAD
AUTHOR
SUBJECT


SEARCH

IPS HOME


    [Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

    Re: iSCSI: DLB-T.16 (resource requirement for connection reinstatement)



    David,
    
    "Connection reinstatement" is related to ErrorRecoveryLevel only in 
    one case (described below).  Even when no tasks were active (so 
    no within-command/within-connection/connection recovery), connection 
    reinstatement still is meaningful - simply the process of substituting one 
    iSCSI connection with a new one bearing the same CID.
    
    To summarize level of requirement for connection reinstatement wrt 
    ErrorRecoveryLevel -
    
    ErrorRecoveryLevel                              target's level of requirement
        0                                                             SHOULD
        1                                                             SHOULD
        2 (multi-conn sessions supported)            SHOULD
        2 (only single-conn sessions)                    MUST
    
    [ The reason for the MUST in the last row is: While the target isn't aware of a
       connection failure, the initiator may have seen the connection fail and would 
       like to reassign tasks to a new instance of the same CID (i.e. connection 
       reinstatement) - target had already promised to support reassignment and so
       MUST allow this as a pre-requisite to reassignment. ]
    
    My suggestion was to require SHOULD in the first 3 rows.
    
    Thanks.
    --
    Mallikarjun
    
    Mallikarjun Chadalapaka
    Networked Storage Architecture
    Network Storage Solutions
    Hewlett-Packard MS 5668 
    Roseville CA 95747
    cbm@rose.hp.com
    
    ----- Original Message ----- 
    From: <Black_David@emc.com>
    To: <cbm@rose.hp.com>; <ips@ece.cmu.edu>
    Sent: Wednesday, July 10, 2002 12:56 AM
    Subject: RE: iSCSI: DLB-T.16 (resource requirement for connection reinstatement)
    
    
    > Mallikarjun,
    > 
    > I think this ought to be related to ErrorRecoveryLevel - it looks like
    > you're suggesting that connection reinstatement SHOULD be supported
    > at ErrorRecoveryLevel 0 - at 1 and above, I believe it's already a
    > MUST.  I don't have a problem with that, does anyone else?
    > 
    > Thanks,
    > --David
    > 
    > > -----Original Message-----
    > > From: Mallikarjun C. [mailto:cbm@rose.hp.com]
    > > Sent: Tuesday, July 09, 2002 2:52 PM
    > > To: ips@ece.cmu.edu
    > > Subject: iSCSI: DLB-T.16 (resource requirement for connection
    > > reinstatement)
    > > 
    > > 
    > > > [T.16] 4.3.4  Connection reinstatement
    > > > 
    > > >    Targets should support opening a second connection even 
    > > >    when they do not support multiple connections in Full Feature Phase. 
    > > > 
    > > > Looks like that ought to be an upper case "SHOULD".  I think this needs
    > > > further discussion.  Section 5.2 appears to use a lower case "must"
    > > > for this:
    > > 
    > > Let's be careful here - section 5.2 discusses the generic case of
    > "connection 
    > > cleanup" - that includes both implicit and explicit logouts - whereas
    > "connection 
    > > reinstatement" is just implicit logout.
    > > 
    > > While "connection cleanup" does not require additional connection
    > resources
    > > in the general case, a target wanting to support connection reinstatement
    > SHOULD
    > > support *opening* a second connection (though it may reject the Login if
    > it's
    > > for a different CID).
    > > 
    > > I believe that connection reinstatement is a useful functionality that the
    > targets 
    > > SHOULD support, so I agree that the change suggested by David is a
    > reasonable 
    > > one to make in 4.3.4 which only discusses connection reinstatement.
    > > 
    > > Mallikarjun
    > > 
    > 
    
    


Home

Last updated: Wed Jul 10 20:18:53 2002
11252 messages in chronological order