kicked qmail

[Ian Lance Taylor]

Zack Weinberg <zack@codesourcery.com> writes:

> On Mon, Aug 26, 2002 at 07:31:22AM -0700, Ian Lance Taylor wrote:
> > Zack Weinberg <zack@codesourcery.com> writes:
> > 
> > > A side thought, there are a number of rare but extremely CPU and I/O
> > > intensive operations that it might make sense to provide a second
> > > machine for.  For instance, generating a GCC release involves checking
> > > out a complete tree and *building* it.  This has to happen in a
> > > context that has local disk access to the repository trees.  I suspect
> > > that NFS would do, though.  Perhaps the old machine, when
> > > decommisioned, could be moved into this role?
> > 
> > Why do you need local disk access to the repository?  I can't think of
> > any reason why that would be necessary.  Checking out the repository
> > via local disk should not be dramatically faster than checking it out
> > on a LAN, and I can't imagine that the checkout time is a real
> > bottleneck anyhow.
> 
> Joseph Myers explained why (and thank you for the corrections, Joseph).
> I'd only add that I think it might be useful to have a second machine
> with shared access, for periodic tasks like snapshot generation that
> shouldn't bog down the main server.

Hmmm, I read his message as explaining why the process did not local
access to the repository.  I guess I am confused.   Can somebody
explain it again?

Ian