On Fri, 21 Apr 2006 01:01:08 -0700, Walter Marguccio
<[EMAIL PROTECTED]> wrote:
>Yes, IXC102A message is very clear about when to reply DOWN. I think I
read somewhere else (SHARE or IBMMAIN)
>that DOWN must be replied only *after* the system has been reset. What I
am seeing is that delaying
>the reply to IXC102A will leave all the remaining LPARs in the PLEX in
a 'hang' conditions. RMF III, option XD
>(XCF delays) shows this clearly. Obviously there's no reason to delay the
reply to IXC102A, but I would like to automate
>this last step too, and was wondering if you can do it with SFM, as the
book states.
Walter,
When the documentation speaks about "automating" IXC102A, what it
really means is that the system will not issue IXC102A at all, with SFM
set up appropriately in a parallel sysplex. Barbara is correct - in a
parallel sysplex, SFM can isolate the outgoing system by fencing it from
the I/O subsystem, as long as there is a CF having connectivity to both
the outgoing system and some other system in the plex that can initiate
the isolation. You cannot achieve this function in a basic sysplex.
The reason for requiring the outgoing system to be reset is more than
releasing reserves. The remainder of the sysplex must be absolutely sure
that the outgoing system is incapable of further I/O that could affect
resources shared among sysplex systems. Only when this is known - by
operator confirmation of manual system reset when automatic isolation is
not possible - can other systems safely clean up for the outgoing system
by taking over its ENQs, locks, etc.
Bill Neiman
z/OS Development
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO
Search the archives at http://bama.ua.edu/archives/ibm-main.html
