James Lampert wrote:
Tom Liotta wrote:
You might want to look deeper. I've been away from that specific
element for a while and never was that deeply into it; but, IIRC,
the default base LPAR changed from 0 to 1 somewhere around V5R3,
possibly with systems new with V5R4?
But what happens if, say, a 170 (not LPARed, probably not even capable
of it), which reports its LPAR as 00, gets updated to V5R4? Is it still
00, or does it automagically start reporting its LPAR as 01?
Wish I could give an authoritative answer. Best I can offer is that
(1) _my_ 170 sure ain't ever gonna have V5R4 installed and (2) more
appropriately to your situation, I don't recall any machine LPAR
_changing_ to something else.
I put together a fairly useful 'hack' that allows us to issue unique
LPP licenses either for a physical machine (all LPARs within a
serial number) or for a particular LPAR. So far, we haven't seen any
cases where i5/OS (nor OS/400) reported a license being invalidated
due to upgrades.
But, it's only been in the wild for a couple years. Maybe that's
long enough to gauge. It seems better to me than assuming LPAR 0 for
an absolute default, at least from what's known at the moment.
In our case, we don't need to use a default LPAR# because I can
determine if the license was _intended_ for an LPAR. If intended for
a non-LPAR customer (because the customer requested one), it's
reflected in what I get back from the licensing APIs. I don't check
the LPAR# unless the API indicates that I should; and the
materialized LPAR then should correlate with what the licensing API
returned.
I figure our licenses now are as secure as IBM's because we use
their APIs to generate, enter, retrieve and verify/test; and if
anything goes really wrong, I mostly get my bosses looking at IBM
rather than me. And if nothing else, switching to the IBM scheme has
our customers seeing familiar IBM panels and commands in place of
what we have been working to get rid of. (Not building that stuff
into our latest products was almost worth the switch all by itself.)
Not much consolation for you at this point, I'm sure. But it's a
degree of evidence for you that a given system image should remain
reasonably reliable. It's been that way for us. (Then again, we take
advantage of the 'grace period' facility too. An LPAR change
wouldn't raise alarms much for us. Nevertheless, I suspect I'd be
dragged in by my boss to explain.)
If I had better info for you, I'd pass it on. Looks like Vern had
some solid info. Everything he wrote seemed to match my memory.
Tom Liotta
midrange.com
