Our client platform is mutualizing many different projects (HR, Finances, Billing) under BO XI 3.1 FP1.3 on AIX.
Before considering an update of BOXI on the server, the current process is a global non regression of all project with the new version.
I understand that this needs to be done for a migration from 2 major versions like going from XI R2 to XI 3.1…
Does this kind of procedure (global non regression test) has to be followed before applying a Fix Pack ? or SP2 on XI 3.1 ?
What are SAP Best Practices or recommandations about this ? Does applying a Fix Pack means full non regression test of every project ?
Do you know of any documentation ?
Do Fix Packs only includes known bug fixes ? What is the risk of applying a Fix Pack ?
Well I remember Service pack for XI R2, like SP3 or even first release of SP2 : there were lots of problem with these releases…
I also encountered problems with Fixpack breaking previous fixes from earlier FP…
But I’haven’t heard of such trouble with XI 3.1 Fixpacks … anyone has ?
My last two XI upgrades (from SP1/MHF1/CHF15 to SP3.4, and from SP3.4 to XI3.1 SP2) both caused data within reports to change. So, yes, I would recommend regression testing for any upgrade.
Again you’re talking about a major version upgrade…
I’m just talking about applying a Fixpack ie : XI 3.1 FP1.3 to FP 1.8.
Should a full non regression cycle be done for a tiny Fixpack upgrade ?
And what about applying service pack 2 to XI 3.1 FP1.3 ?
Let’s say our server has stability issue when charged (AIX 3.1 FP1.3)… Fixed issue list in FP 1.8 states that at least 2 major bugs could be related to stability in such condition…
Should a full non regression test be done just for a Fixpack ?
Everybody does so ? Anyone didn’t ? What is SAP/BO saying ?
My quote referred to BO XI R2 in early 2008 when updates caused lots of problems…
Since then, it seems things are going much better with less regression… am I wrong ?
I never meant I’d be confident in upgrading a production system without at least some testings
But does going from BO XI 3.1 FP 1.3 to FP 1.8 to gain in stability means performing the same test procedures as going from BO XI R2 to XI R3.1 ?
I’m talking about hundreds of tests step for a douzain of projects and calling users and dev team to restest their report just for a Fixpack !
Am I wrong or to naive believing a few representative comparaison of sample reports after refreshes on 2 servers before and after Fixpack on same datasets should be enough ?
And what about Fixpack uninstall procedure ? Is it safe ? Does it needs to be fully tested too
I’ve read you couldn’t go from SP2 back to 3.1 if universe or report have been modified… what about Fixpacks undo ?
Of course not! Jumping to a new release should be tested throughly.
I’m just saying that it is my understanding that BO do not regression test fix packs.
It is also my understanding that you only install them if you have to. I understand, completely, why you want to apply this fix to your set up.
As I said earlier, they haven’t had many chances to mess them up yet, there have not been that many fix packs ;).
Hopefully SAP will have a good influence on BO software development and they will improve.
Well, I’m sure you have a Development system - lets hope you don’t have to find out.
I have had some terrible experiences, over the years, trying to remove Business Objects components and have had to rebuild servers on a number of occassions.
This is correct.
As for fixpacks I think people have had mixed results when rolling back, I can’t speak for 3.0.
My basic message is you can’t generalise, with regards to this, you may have a problem, you may not, simple as :).
Keep in mind that a legitimate fix may negatively affect your reports. That is what happened to us when we applied SP3.4 in XIr2 – we had (unintentionally) worked around a bug in the calculation engine, but we had the correct results. When the fix was applied, our numbers changed.
As far as what to test, my recommendation would be to focus on the key, or representative sample reports. We have a set of critical reports that are tested in their entirety for every patch. Other, less sensitive reports are not tested as rigorously.
We haven’t had a single SP/FP applied that didn’t result in new problems.
Best Practices in the IT world is to test any changes before applying to your production environment. We have all learned in the past during our testing that there are situation when a “fix” will create an issue while correcting the intended problem. Business Objects does not seem to be QA testing the fixes thouroughly and every FixPack and ServicePack should be treated as you would a “Beta Test”.
Business Objects just released Fix Pack 2.2 which requires SP2. FixPack2.2 includes prior Fix Pack 2.1 and 2.2 include fix pack 1.6, 1.7 and 1.8. Fix pack 1.9 would be included in FP 2.3.
Yes, it is expensive to do complete regression testing; maybe that’s why BusObj/SAP skimps on their QA. Even if you read on some SAP website about an official policy about their QA, I wouldn’t trust it.
(BOB member since 2003-05-26)
(BOB member since 2005-01-06)
.
(BOB member since 2002-08-29)
