Thank you for explanations but I still can not get the point. I see that 
disabling disconnection once and for all is a bad idea because the 
disconnection is normal routine for coda. But  I still do not understand 
why having the coda client on a machine, where one of a set of 
replicated servers resides, is any worse than usual.

> if the system is
>trashing, it will affect the client just as much as the server.So the
>client might be blocked while swapping and not see the response from the
>second server fast enough.
>
>  
>
But here the situation jast as bad as with client and server which are 
on different and heavily loaded machines.  The client is already 
disconnected from the local server and the interaction with client and 
the next replicated server proceeds the same way as if the local server 
was absent from the beggining, isn't it? Yes,  the local server does put 
a load on the machine, but suppose we have another application (instead 
of the local coda-server) which is greedy for memory and makes 
coda-client swap a lot?

Thank you,
M.Kondrin