I recently reconfigured some replication from a single node to 2 
subscribers.
One of the subscribers worked very well and is now in sync.
One of the subscribers has some network connectivity issues and has been 
working in fits and starts (never actually completing the subscription 
of the first set).
So today, I removed the node from the replication cluster via pg-admin 
(mistake?).

However, there now appears to be a LOT of stale/old entries in sl_event, 
log_1, and log_2 (and wherever else) representing that old node.
I did a delete cascade on the replication schema on that dead node in 
preparation for recreating the database from scratch, but am at a loss 
as to how to deal with this 'bloat' on the master node side.


How can I clean this up...and what SHOULD I have done to drop that node 
properly?