> My questions are: a) Where am I going wrong ?

What is wrong is that you need a much fuller set of paths and listen
settings.  If there are 3 nodes, the paths and listener settings need to
look more like the following:

cluster name = transtest;
 node 1 admin conninfo='host=host1 dbname=transtest user=postgres
port=5432 password=postgres';
 node 2 admin conninfo='host=host2 dbname=transtest user=postgres
port=5432 password=postgres';
 node 3 admin conninfo='host=host3 dbname=transtest user=postgres
port=5432 password=postgres';

      store path (server = 1, client = 2, conninfo = 'host=host1
dbname=transtest user=postgres port=5432 password=postgres');
      store path (server = 2, client = 1, conninfo = 'host=host2
dbname=transtest user=postgres port=5432 password=postgres');
      store path (server = 1, client = 3, conninfo = 'host=host1
dbname=transtest user=postgres port=5432 password=postgres');
      store path (server = 3, client = 1, conninfo = 'host=host3
dbname=transtest user=postgres port=5432 password=postgres');
      store path (server = 2, client = 3, conninfo = 'host=host2
dbname=transtest user=postgres port=5432 password=postgres');
      store path (server = 3, client = 2, conninfo = 'host=host3
dbname=transtest user=postgres port=5432 password=postgres');

## and listeners...

      store listen (origin = 1, receiver = 2, provider = 1);
      store listen (origin = 1, receiver = 3, provider = 1);
      store listen (origin = 2, receiver = 1, provider = 2);
      store listen (origin = 2, receiver = 3, provider = 1);
      store listen (origin = 3, receiver = 1, provider = 3);
      store listen (origin = 3, receiver = 2, provider = 1);

You're missing the communications between nodes 2 and 3, and that's
probably what has gone wrong.

If you take a look at the tables sl_path and sl_listen, on all three
nodes, you'll probably find that instead of having 6 entries in each table
(which is what you need), node 1 only has 4, and nodes 2 and 3 may only
have 2.  There should be 6 in each.

Even though nodes 2 and 3 aren't directly speaking, they still need to be
on "speaking terms" in order that SYNCs make it back and forth so all the
nodes can know when to purge out sl_log_? entries.

Revise the paths and listens and see if that takes you some steps ahead...