[pca] PCA Forum Link - And question about PCA

Martin Paul martin at par.univie.ac.at
Wed Nov 26 10:00:53 CET 2008


Jack VAN GORKUM wrote:
> First, thanks for the PCA application. We have been using PCA for some
> time now, and like the results (especially the HTML formatted reports).

Thanks for the feedback!

> Can you send me the link to the actual forum (if there is one) as I
> cannot find it.

There is none. Personally, I never got the hang of web forums - I hate 
wading through 10 pages in a thread looking for the messages which I 
haven't read yet. There's an archive for the mailing list, though, if 
you're looking for past messages:

   http://www.mail-archive.com/pca@lists.univie.ac.at/

> We noticed several patches which show up on PCA that do not show up with
> Sun's Update Manager. It turns out the more accurate report is PCA's
> information.

Yes. Theoretically, Sun's UM should be better, as it has more 
information about the patches. Practically, it still has issues with 
false positives (patches which don't apply to a system) and false 
negatives (it doesn't show some patches, even security, giving you a 
false sense of safety) all the time.

> When we contacted Sun, they informed us the following;

I'ma amazed about the reply you got. You must be an important customer.

> 	"PCA utilises the patchdiag.xref file which is included with
> patches. However, this file is dependent on being accurate in the first
> place and (cannot be guaranteed to continue to be updated.) 

Of course pca depends on the information provided by Sun in the 
patchdiag.xref file. It isn't the only source it uses though - over the 
time I've collected a big amount of extra rules for problematic patches 
in pca itself.

For a long time, I was hesitating to add such "hacks", but at the end it 
turned out to be a good thing. As more and more people started to use 
pca, it was being run on a lot of different systems with exotic 
installations and package sets. So sooner or later, if a patch shows up 
wrongly, it got/gets reported to me and I add a workaround to pca for 
that patch, usually on the same day. Over time, it got more and more 
accurate like that.

The problem with UM's "patch realisations" is that your chance to get 
one added or fixed is very low, and it usually takes weeks or months. 
During that time, thousands of admins trip over the same problem and 
waste a lot of time looking at issues which could have been solved 
within minutes a long time ago.

> Are there plans to modify PCA to use whatever patch realization method
> (or info) Sun will guarantee to keep up to date ?

The problem with using the current patch realizations is that they 
consist of (compiled) java code and are not documented, so there's no 
way to use them with an external program.

The history of Sun's announcements of patchdiag.xref being removed is 
long. There is no guarantee that the file will continue to be provided, 
at least I can't give you that. If Sun stops providing the xref file 
tomorrow, pca is dead as well. I'm pretty optimistic that this won't 
happen soon, though. There are people at Sun which are fans of PCA.

Martin.



More information about the pca mailing list