Don O'Malley wrote:
Yeah, I do realize that it may seem like a huge difference to you as
> you are very close to patching related issues, but unfortunately many
> customers if given the choice, would prefer to always install the
> latest revision of a particular patch. I guess the logic is that newer
> is better!
Actually, I'm more like "many customers" in that regard, usually using
only the most recent revisions of patches. Most patching is done with
"pca -i" (install all recent, missing patches), minus a few machines
which only ever get patched if I experience a problem.
IMHO, this all-or-nothing approach has a big advantage: Both put the
system in a state which has well been tested by Sun (a clean release or
a fully patched system), which might not be true about any other given
combination of installed and uninstalled patches.
Still, I understand that some people want the minimum set of patches
which Sun recommends to fix the most urgent problems. The Sun Alert
cluster is probably very close to that definition.
> To quantify this, I've examined the contents of the Solaris 10 Sparc
> Recommended Patch Cluster (which contains 155 patches) and there were
> only 3 patches in the cluster which did not have the 'R' flag in the
> patchdiag.xref file.
Ah, great, I didn't know that there is such a close match already. I
remember that it wasn't like that when I checked the last time, but I've
seen the 'R' coming and going in big waves in the past.
> These patches correspond to the following patchdiag entries:
And that's just a well-known patch anomaly (122660 and its requirements,
where an obsoleted patch is required under some circumstances), which is
already taken care of by pca as well.
So at least for Solaris 10 one can assume that installing the
Recommended Patch Cluster and running "pca -i missingr" will get you to
(about) the same patch state.
> Please note that the SunAlert report contains a list of all patches
> (not just patches for Solaris) which address a SunAlert issue.
I had thought about using SApatches.txt within pca, but ended up with
the usual issue with non-recent patch revisions - patchdiag.xref doesn't
have the information about those, making it impossible to resolve patch
requirements.
Thanks for all the information!
Martin.
