I spoke w/ Peter Braam one of the professors who worked on Coda, he said the project is dead. refer to Intermezzo
*** Coda isn't actually nailed to its perch - just "stable", so research is on follow ons based on the same underlying principles.
I checked inter-mezzo but most of the docs, like most of the Coda docs, are describing specific implementation issues for unix kernels and file systems.
The inter-mezzo research papers are also still at the Coda docs reference I gave before:
The abstract protocol that enables optimistic conflict resolution with disconnected clients and partitioned networks is not described in most of these docs but just referenced.
That, and not the actual Coda or Inter-mezzo implementations is what I believe could be highly relevant to ZODB replication.
One online paper listed there that does describe it is Lee's thesis:
Please take a close look.