> Date: Wed, 26 Nov 1997 09:18:23 -0800
> John J Archer writes:
> >
> > I just got off the phone w/ HPRC and the word from the lab is
> > that there are several patches on the tape that will not install
> > via Patch/iX. For my system (at least) there were 9. The most
> > obvious was MPEJXR6 (the ci enhancement patch). I'm instructed to
> > use AUTOPAT or some other option like HPINSTAL.
I have talked to John, and the HPRC engineers involved. There are a
couple of subtle "misleading" phrases in the message that John
posted.
(1) There are 9 patches on the Express 3 tape that will not "qualify
naturally". Saying a patch will not "qualify naturally" is different
that saying it will not install. Patch/iX should be able to install
all of these patches.
(2) The HPRC is recommending that customers use AUTOINST (not
AUTOPAT). AUTOPAT is only capable of installing reactive patches, not
PowerPatch tapes (patches).
I have been investigating this issue and I have some additional
comments below. Specifically, I don't agree with the blanket
recommendation to use AUTOINST instead of Patch/iX. The differences
between AUTOINST and Patch/iX with respect to this particular issue
are:
(a) AUTOINST does not differentiate between "enhancement" type
patches and "regular" patches. Patch/iX does. In Patch/iX,
"enhancement" type patches do not automatically or naturally qualify.
To install "enhancement" type patches using Patch/iX, the customer
must "force" the patch -- a standard Patch/iX "feature".
NOTE: this is the way Patch/iX is designed to work, it is not a bug.
(b) AUTOINST will always install all of the patches that qualify.
Patch/iX, by default, will install all of the patches that qualify
"naturally", but will always give the enduser the option to "veto"
patches that he/she does not want, or to "force" patches that did not
qualify naturally that the system manager does want/need.
This issue does not warrant the recomendation to use AUTOINST instead
of Patch/iX by itself.
[John look at the NOTE below concerning the more complicated issues
related to the order of events and reactive patches.]
Mark Bixby wrote:
> This sounds like HPRC unfamiliarity with Patch/iX.
This is both true and false. In actuality, some key members of the
HPRC did a very good job of diagnosing and debugging this "issue".
One of them wrote a very detailed explanation of the "issue"
including recomendations. This message was forwarded throughout the
HPRC and in the case of John Archer, the online engineer that took
his call had read the explanation and reponded appropriately
according to the recomendations in that message.
The reason I point this out is because it is good news. The HPRC is
developing better working knowledge of the Patch/iX tool. In the case
of John's call there are some additional complicating factors (becaue
he was also installing reactive patches and he actually did the
update before he noticed that the 9 patches were not installed) --
these are discussed below.
The only issue, as I see it, is that I think the recomendation to use
AUTOINST instad of Patch/iX is "overkill", and there are a few
terminology issues. Specifially, the idea that not qualifying means
that the patch won't install is not correct. The original HPRC memo
was clear on this, but some of the message was lost in translation.
NOTE: the situation gets a little more complicated depending on what
order you did/do things in. The complications show up if you also
install other reactive patches at the same time that you install
Express 3.
Consider this scenario (which I believe corresponds to John's full
scenario):
* If you use Patch/iX to install Express 3
* AND you don't notice that the 9 patches did not qualify
* AND at the same time you install additional reactive patches,
* AND then you go back through and try to force the original 9
patches with Patch/iX
you may have other problems due to patch dependency/supercedes issues
between the 9 patches on the Express 3 tape and the reactive patches
you installed the first time through.
[John this is the problem you are running into & the reason why the
patch MPEJXR6 had erors when you tried to force it -- at least I
think it is :-)]
If you use AUTOINST to re-aply the Express 3 tape, then you will
install all the patches on the Express 3 tape - PROBABLY! What does
probably mean? Well, it really depends on what the other reactive
patches were. It is possible that some of the patches on the Express
3 tape will not qualify for more legitimate reasons now that you have
installed additional reactive patches. This includes, but is not
limited to, the 9 patches that did not qualify originally.
This is actually the most complicated case. The "right thing to do"
is to "take inventory" of which patches are actually installed on
your system, and which ones you want but are not installed. This can
be done by using Patch/iX and the "View Patches" screen selecting the
"Installed Patches" filter. Alternatively this can be done the hard
way by looking at the various Patch/iX log files (PMLOG and PMAUD)
provided you haven't run Patch/iX again [in your case John, you
have].
Once you have a list of what is actually installed on your system,
and what is missing, you can request copies of all the patches you
are missing and install them as Reactive Patches using Patch/iX.
NOTE: some of the missing patches will have been superceeded so you
will actually get different/newer patches if you do this. Your HPRC
person should be able to help you with this.
[back to Mark's message:]
> After Patch/iX qualifies the patches, scan the list for anything that didn't
> qualify. Highlight the patch with the cursor keys, and press Enter. There
> should be a function key called something like "Next filter view" that you
> can press to view various information screens about the patch. One of these
> screens will tell you which component(s) of the patch had a problem and why.
This is correct. In the case of all 9 patches on the Express 3
PowerPatch tape the "reason" should state that the patch did not
qualify because it was an "enhancement" patch. It it give you any
other reason, especially anything related to checksums or the patch
being superceeded then you should not force the patch.
> In my case, CATALOG.PUB.SYS was flagged with a checksum mismatch because I
> had modified the catalog to add some error messages. Not a problem, so I
> "forced" the patch to qualify anyway. I could fix up CATALOG again after
> I had installed Express3.
This is an excellent example of:
(a) a Patch/iX user that understand exactly what is going on and how
to use the tool effectively
(b) a type of local modification that would cause Patch/iX to think
there is a problem. I know that other customers also modify the
catalog, they will see this. The workaround in this case is easy and
clear. In fact, customers who modify the catalog are used to having
to reaply there catalog changes after installing any patch that
replaces catalog.pub.sys with any of the existing patch tools.
> Note that most enhancement patches fail to naturally qualify just because
> they are enhancements. You must explicitly force these if you want the new
> functionality.
Actuall "all" enhancement patches are supposed to fail to naturally
qualify -- by design. If only "most" of them are failing then I would
consider that to be a bug (but I have not heard of that happening).
> Never force anything other than an enhancement or local-mods-cause-mismatch
> patch without talking to the HPRC first!
This is very good advice. In the case of the patch not qualifying
because of what Mark calls a "local-mods-cause-mismatch" type
problem, I would suggest that you dicuss this with the HPRC if you
have any doubt as to how to resolve the problem. If it is clear to
you, then go ahead and force the patch.
Scott McClellan
___ ___ _________ Hewlett-Packard
/_ /| /_ /| /_______ /| Commercial Systems Division
|##| | ##| ||########| | 19447 Pruneridge Ave
|##| |_##| ||##| |_##| | Cupertino, CA
|##|/__##| ||##|/__##| |
|########| ||########|/ E Mail: [log in to unmask]
|##| | ##| ||##| | Phone : (919) 969-7870
|##| | ##| ||##| | Fax : (919) 969-7871
|##| | ##|/ |##|/
-- -- -- Voice Mail : 447-6067
|
