Using distcc, ccache for twin servers

[depontius]

I have a pair of (ancient) machines of roughly equal capability that I use for home servers, one backup, one active. They are sufficiently old that it's rather painfully slow at software update time.

I've read of distcc and ccache, but it doesn't sound quite like what I want. Since they're identically configured, both gentoo-hardened, both P-III Coppermine, (one a Celeron) I'd like them both to share in the task of compilation, each using what the other has done. That sounds almost like running distcc and ccache, but having the ccache space on shared nfs. Each machine would pick up pieces of compilation from distcc, putting it in the shared ccache. Each machine would pick up pieces, whether self or other compiled for linking and installation, etc. Kick off both "emerge -uvDN world" at the same time, both with distcc, both with "-j2", because each would start on thread of its own and one thread for the other system.

Does this make sense?
Has anyone done something like this already?
Is it horribly fraught with peril?

Obviously I could use one system to build binaries for both, using distcc, then install the binaries on the other. But if possible I'd prefer a peer-type arrangement that could even work identically if one machine were down, just slower.
_________________
.sigs waste space and bandwidth

[SeaTiger]

Simple way:
-- use distcc
-- emerge with buildpkg
-- machine A for emerge source and compile job(still distribute compilation job to machine B via distcc) and bin-package server
-- machine B emerge bin package from machine A

Complicated way:
-- hmm ... still thinking ...

[depontius]

I'd kind of figured that was the simple way. I was hoping to come up with something "symmetric" by putting the ccache on a shared nfs, but I can see how it might cause problems.
_________________
.sigs waste space and bandwidth