SORT BY:

LIST ORDER
THREAD
AUTHOR
SUBJECT


SEARCH

IPS HOME


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

    RE: iSCSI: Need to Kill Session from Surviving Node



    
    OK, let me see if I can stretch it a bit more.  The key to what you said,
    is for the Initiator needs to be sending something to the target, not just
    waiting.  So if the Initiator always sent a Nop-Out after some period of
    time of inactivity from the target, (hopefully a lot smaller time then the
    TCP/IP Time Out) then the Fail-Over node would be able to cause the failure
    reasonably quickly by returning ICMP Destination Unreachable.
    
    If I have it right, it seems straight forward, but of course you need to
    take over the IP Address.
    
    
    .
    .
    .
    John L. Hufferd
    Senior Technical Staff Member (STSM)
    IBM/SSG San Jose Ca
    (408) 256-0403, Tie: 276-0403,  eFax: (408) 904-4688
    Internet address: hufferd@us.ibm.com
    
    
    Black_David@emc.com@ece.cmu.edu on 03/15/2001 06:26:44 PM
    
    Sent by:  owner-ips@ece.cmu.edu
    
    
    To:   John Hufferd/San Jose/IBM@IBMUS, Robert.Elliott@compaq.com
    cc:   ips@ece.cmu.edu
    Subject:  RE: iSCSI: Need to Kill Session from Surviving Node
    
    
    
    > What I was trying to get at is ---  trying to enable the
    > Surviving Node (which has no TCP/IP state from the Failing
    > node) to get the Initiator, which had a previous Session with
    > the Failing Node to take actions right away, and not wait
    > for its TCP/IP connections with the Failed Node to Time Out.
    > (This requires some sort of a message from the Surviving
    > node back to an Initiator.) This is not quite what we have
    > defined into iSCSI.
    
    Since this is a TCP problem (get the connection to close),
    iSCSI may not need to do anything ... what about failing over
    the IP address and responding with ICMP Destination Unreachable
    to anything that arrives on that failed-over IP address?
    Implementations of this probably need to support multiple
    IP addresses on a single network interface, which is not a
    big deal.
    
    --David
    ---------------------------------------------------
    David L. Black, Senior Technologist
    EMC Corporation, 42 South St., Hopkinton, MA  01748
    +1 (508) 435-1000 x75140     FAX: +1 (508) 497-8500
    black_david@emc.com       Mobile: +1 (978) 394-7754
    ---------------------------------------------------
    
    
    
    
    


Home

Last updated: Tue Sep 04 01:05:18 2001
6315 messages in chronological order