[SlugLUG] New Member Here!

[cerise at armory.com]

On Wed, Sep 20, 2006 at 05:14:44PM -0700, Ignacio Solis wrote:
> cerise at armory.com wrote:
> > Failing anything else, you're doing the kernel team a favor by staying up to
> > date.  It's the least you could do for them.
> 
> ?? The kernel team releases kernels so that people use them. It's not my fault
> that some of the releases are more successful than others. Using one means it
> was a very good release.

   You missed the point completely.  Spend a little bit of time on, say, the
linux-smp mailing list and you'll see that about half the traffic is people
bug reporting things in bygone versions of the kernel.  By the time any of
those things are bug reported, the code which caused the bug has most likely
been modified in a number of different ways.
   You're doing the kernel team a favor by using the current version of the
kernel.  It ensures that if you do run into a problem, it's on code that
still exists.  It also ensures that you're getting the very latest of their
labors.
 
> But you are not a server. 

   Actually, my main system is a server.  l.armory.com.

> The SlugLUG server probably has no need to update to
> 2.6.18. There might be essential parts, these might get back ported. Or maybe
> they'll be so intrinsic as to force an upgrade in the Debian stable kernel.
> But, for example: who needs a patch that:
>   "[Fixes] performance degradation introduced in 2.6.17"
> ... oh yes, people that upgraded to 2.6.17.

   And the same people who upgraded to 2.6.17 would then upgrade to 2.6.18.  I'm
not really sure what you're getting at.
 
> Don't get me wrong, I'm not against upgrading the kernel. I'm just saying that
> some servers just don't need the latest cutting edge kernel.

   And you might well have upgraded to the equivalent of 2.6.17 ensuring that
your shiny, new linux system might well not be doing all it could or should.
 
> Most of these problems are dealt with by the Debian kernel team. So you don't
> really need to worry about them.

   My, aren't you a trusting soul!  For one thing, these problems aren't merely soluble by the Debian kernel folk, but also the application folk.  And 
considering the lag on Debian's acceptance of X.org's version of X and their
handling of DRI before then.
   I trust myself as a capable sysadmin to solve the problem at the application
and kernel levels much faster than I'd trust teams of people mired in politics.
 
> The SlugLUG server has no need to care about PLIP, and much less about ADFS.
> Who uses the Acorn Advanced Disc Filing System anyway?

   I do, as a matter of fact.

> Tc and family could be useful at some point, but we don't use it, and the
> debian kernel comes with a default config that works in most cases. Most
> schedulers are modules, so just load the ones that you want.

   My point is that in adopting odd things, I wouldn't trust the release teams
to have tested them against their back ports.  I tend to be ahead of the curve
on a lot of things.
   Again, I trust myself a lot more than I trust teams of people concentrated
on the general case and -- again -- mired in politics.
 
-Phil/CERisE