Re: Recovering from a hosed up CUME install - RESOLVED

[Pete Helgren <Pete@xxxxxxxxxx>]

I'll reply for the archives.

I took several approaches to get to the bottom of the issue.  Basically 
I decided to follow Pete M.'s suggestion to write a CL program that 
would read through the PTF's that did not apply and individually apply 
them and I used Rob's suggestion to set the IPL Attribute STRRSTD(*YES) 
[CHGIPLA STRRSTD(*YES) ].  Then I ran the CL and bounced the i, about 5 
times finally, having nothing apply and about a dozen PTF's that weren't 
applying .  I pursued a few and they ended up being truly circular.  So 
i decided just to apply the CUME and bounce and ended up with about 2 
dozen that didn't apply and it seemed like the same 6 or 8 PTF's were at 
the center of all of the issues.

So tonight I slipped the LIC and applied the CUME and Voila! all the 
issues resolved.  I noticed that there was one special handling PTF that 
required an extra IPL and then everything applied smoothly.  The good 
news was that I only needed to load LIC and not OS which kept things 
fast and simple.

You gotta love i5/OS!  Imagine if you had a hosed up install of Windows 
Service Packs: You'd have to blow the whole install away and reload from 
scratch.

Thanks for the help and suggestions.

Pete

Pete Helgren wrote:
> I have a customer that, for the life of me, managed to install a CUME 
> that just never seemed to fully apply (V5R4).  Over a year ago I 
> invested 3 or 4 hours tracking which PTF(s) was(were)  a co-req or a 
> pre-req and eventually was so confused by the circular logic (who's on 
> first, what's on second...) that I have decided to start with a new CUME 
> and apply it *clean*.  The question is, can I *undo* the previous CUME?  
> I *think* the answer is yes, if I can slip the LIC and then load and 
> apply the latest CUME.
>
> Is that the best approach?  Or should I just I just plan to upgrade to 
> 6.1 and leave it at that? I'd rather not do the 6.1 upgrade at the 
> moment so if there is a way to recover from a *bad* CUME install, I'd 
> like to take that approach first.
>
> Pete
>
>