Using slony1 1.2.15, if I try to subscribe a set and the receiver node
generates an error so that Slony retries, even after fixing the issue
that caused the failure, the subscription will never succeed.  The
slon daemon for the receiver node reports "table id <N> has already
been assigned!".

Here's the full log snippet:

2009-06-18 17:19:11.325322500 DEBUG2 remoteWorkerThread_1: all tables
for set 2 found on subscriber
2009-06-18 17:19:11.328928500 DEBUG2 remoteWorkerThread_1: copy table
"public"."customer"
2009-06-18 17:19:11.330919500 DEBUG3 remoteWorkerThread_1: table
"public"."customer" does not require Slony-I serial key
2009-06-18 17:19:11.383551500 ERROR  remoteWorkerThread_1: "select
"_customer".setAddTable_int(2, 42, '"public"."customer"',
'pk_customer', 'Table public.customer with primary key'); "
PGRES_FATAL_ERROR ERROR:  Slony-I: setAddTable_int: table id 42 has
already been assigned!
2009-06-18 17:19:11.383701500 WARN   remoteWorkerThread_1: data copy
for set 2 failed - sleep 60 seconds
2009-06-18 17:19:11.384672500 NOTICE:  there is no transaction in progress

To proceed, I need to drop the set, recreate the set and resubscribe
the receiving node.  Not that big a deal, but I swear this used to
work properly in the past...

I know I should upgrade to 1.2.16, but I looked through the release
notes and didn't notice anything that seemed relevant to this issue,
so perhaps it still exists there?

-Dave