From: "Christopher Browne" <cbbrowne at ca.afilias.info>
>   Everyone please test their failover and move_set scenarios thoroughly.

I'm afraid I've still got problems with my 3-node 10-set system.

The setup is 1 -> 2 -> 3.

On switchover of the origin of the 10 sets from node 1 to node 2, node 3 
shows two symptoms:

a)  An "ERROR:  duplicate key violates unique constraint..." on one of its 
data tables (seemingly random)

b) Slon dies 10 times (presumably one for each set), restarting 10 secs 
later in every case. Once that has passed, replication seemingly returns to 
normal.

I haven't encountered any data loss.

I'll send the node 3 logs to Christopher and Jan.

Julian