Re: a cautionary tale

[David Kreuter]

Title: RE: [IBMVM] a cautionary tale

no not really, the LPAR that the dynamic I/O was issued from sends the command to the processor controller, which updates the HSA. The LPAR used is just the interface to the really real I/O stored up there somewhere. Yes I agree it was like a power supply failure on a disk drive.  The recommendation is always to use one LPAR for dynamic work. What happened is that someone did a delete i/o instead of a modify i/o.

David


-----Original Message-----
From: The IBM z/VM Operating System on behalf of Nick Laflamme
Sent: Tue 10/10/2006 8:33 AM
To: IBMVM@LISTSERV.UARK.EDU
Subject: Re: [IBMVM] a cautionary tale

David Kreuter wrote:
> Cynic that I am I still think dumb luck played a part here; the linux
> admin thinks otherwise.  In any case linux on the mainframe under 
> z/VM continues to astound and delight.

Deleting devices so a guest can't see them seems analogous to me to a
power supply failing on a disk drive controller or a disk array: the
operating system (and applications) ought to have some way to recover
from Bad Things That Happen, so matter how rare such events are supposed
to be.

I'm surprised, though, that deleting devices from one LPAR would muck up
other LPARs. I take it the dynamic I/O support propagates the "delete"
intent all the way back to the hardware and into other LPARs? That seems
rather "aggressive."

> David Kreuter

Nick