One thing with installing 'everything': you have got to patch everything
too, or keep track of the patches you don't need because you're not
using the package.   Our patch status system has the advantage of
highlighting packages to be patched on systems so that we can think
about the need for them being installed in the first place.  A few show
up now and again, and we can delete them without patching.

We tend to do a base installation of the 'Server' config with tailored
additions/deletions per system type.  Means we can say no when users ask
for browsers on the batch workers ;-).

Martin 'you want what installed?' Bly
RAL Tier1 Systems Team.


> -----Original Message-----
> From: [log in to unmask] 
> [mailto:[log in to unmask]] On 
> Behalf Of Miles O'Neal
> Sent: 10 March 2006 16:23
> To: Scientific Linux Users
> Subject: Re: packages for cluster nodes
> 
> Connie Sieh said...
> 
> |And since they did ASK for everything we should not have to 
> "save people 
> |from themselves".
> 
> On the one hand, I completely agree with you.
> 
> On the other hand, installing everything shouldn't
> mean turnoing on nearly so much (by default).
> That's not your fault; it comes that way from The
> Upstream Vendor[1].  But it's still obnoxious.
> With each major release, I end up spending several
> hours trying to figure out how much of the new
> stuff I should turn off.
> 
> -Miles
> 
> 
> [1] Perhaps we should just abbreviate this as TUV!
>