[Gllug] Production system - Linux 2.4.24, LVM and cciss

[John Hearns]

On Mon, 12 Jan 2004, Rickey Costas wrote:

> Martin A. Brooks wrote:
> 
> 
> > 
> > It's just one of those things.  These systems were all here before I was
> > and, as they're production machines, it's not practical or appropriate
> > to change them.  Not broken, don't fix.
> 
> Surely the work to maintain them is enough justification to standardise 
> ? You have to fix security issues as soon as they come out dont you ? Or 
> spend time installing funky configuration management programs.....
> 
> 
> > With sensible package management, no.  I'm actually quite lucky here in
> > that I've set up cfengine, it does 90% of the work for us.
> 
> cfengine looks prety funky, but also like one of those things thats a 
> solution looking for a real problem ! 
I don't agree.
There are many smart people looking into methods of automatically 
configuring and managing large, heterogenous arrays of machines.
The one I'm more familiar with is LCFG http://www.lcfg.org
The LHC grid project at CERN is evolving an XML based schema to describe
the configuration of machines.


I've never used it so I could be 
> way off.  Maybe I should have a better look at it in all the spare time 
> I've got caused by all my machines being the same :-)
Well then you are lucky.
There are other institutes and industries out there which have a big mix
of machines. Even though they're Linux, they may do different jobs,
such as compute servers, dedicated build machines, login hosts, web 
servers...  
That's where things like lcfg and cfengine come in.



-- 
Gllug mailing list  -  Gllug at gllug.org.uk
http://lists.gllug.org.uk/mailman/listinfo/gllug