[geeks] Suprise - ext. USB HD "Just Worked" under Solaris 10

[velociraptor]

On 11/6/06, Charles Shannon Hendrix <shannon at widomaker.com> wrote:
> > Application of the latest patch cluster for Solaris 9 using a custom
> > script that weeds out non-applicable patches for the given system.
> > E.g. non-applicable hardware drivers, arch patches, etc.
>
> How do you do this?
>
> I never was very successful at automating either installs or patches for
> Solaris.

It's a home grown script from $ork.  I don't see the point for it,
given the latest stuff available for patching, assuming you have
contract support.  I don't know if smpatch works with just a basic
sunsolve account (e.g. no contract connected to it) or not.  Anyone
care to test/comment?

If you are working for a paranoid organization or one that needs to
save bandwidth, and have contract support, you can get a patch server
proxy software from Sun so that you can download updates to a central
repository in your own organization.  This also allows you to keep
your boxes at the same patch revision in the case where it would take
you longer than 2 weeks to patch everything.  (Sun releases patches
2x/month, iirc.)

> Basically I hated how it was set up, but I really do need to learn.
>
> Most admin books really just repeat the man pages or WWW dox on this.

Jumpstart is very, very simple to set up unless you want to do complex
stuff pre/post install with begin/end scripts.  However, jumpstart
simply installs the same way you would if using a CD, e.g. adding each
software cluster individually.  Hence, it is slow for anything but the
most recent hardware release.  You are better off using jumpstart +
flars if you have the disk space.

Another thing to look at is "live update", which allows you to
basically build a "dual boot" system to protect yourself from OS
patches.  The advantage being that the second copy of the OS can be
patched without taking the system down.  This leverages the same
principle that your host box used to build flar(s) from has.

I hate to say it, but I find a lot of Solaris admins (at least around
here) whine about the "maintainability" of Solaris, but never bother
to actually keep up to date on the things that Sun has released to
make it easier.  The admins look at the pretty little RH console that
pushes updates, and they like how they don't have to think about it,
so they cave to using Linux on junk commodity hardware.  Of course,
their hardware calls have probably doubled or tripled, but that never
gets factored in since only the client departments (the admins here
support servers and desktops) see the cost increase on the x86
hardware front.  The sys admins said they hadn't seen a Sun failure in
close to a year, other than server hard drives and one server raid
controller--we had two U80's fail in the last 3 weeks, both post power
outages.

I see the RH web console and think about the crappy QA experiences I
had with RH back in '01, and consider the cascade effect of pushing a
bad update to 200 clients.

=Nadine=