Hi,

for information and for what it is worth ("bad practices" so to say).

Despite the great Coda robustness against flaky connectivity
we have seen clients losing functionality when several hundred well connected
clients updated _different_ files on a volume and there was also
an "unreliable" client reading the files.

The server supposedly spent a lot of time breaking callbacks, which forced
the clients to wait long, longer and after some point indefinitely.

Moving the files to different volumes would reduce the impact,
but not cure the cause. Now we are looking forward to callbacks-free Coda...
while trying to distribute the load to many volumes too.

Another annoying problem with multiple writers to the same volume
has been submitted as a bug today.
It seems that this is another source of "unnecessary" or "impossible"
conflicts. Even if our setup is not quite "natural" for Coda,
it is possibly useful for finding hidden bugs :)
Note that the writers share a volume, but not any files.

Regards,
Rune