Hi.  In the Slony FAQ, section "5.2.8.4. Replication Fails - Unique
Constraint Violation" is listed under "hopefully obsolete".

I've just run into this issue, using Slony-I 1.2.21, running against
Postgres 8.4.8.  So looks like this issue is still around.

I had Slony repeatedly trying to insert values on the subscriber that
were rejected with "unique constraint violation".

Same values do not appear on the master -- I guess the master's
database rejected them.  It appears that Slony was attempting to
replicate to the subscriber a transaction that was rejected on the
master?  The description in 5.2.8.4 fits to a "t".  Looks like this
"glitch" is still around.

Aleksey