I'm trying to rebuild a pair of replication/logshipping databases that I
accidentally managed to screw up a few days ago.  (Screwed up the provider
node, so I'm assuming from the instructions I found in my predecessor's
notes that I'm going to have to drop/recreate the subscriber as well.)  I
have a backup from the early on the day of the error (well before I managed
to screw up), so I've got the most recent data available.  However, I'm
running into a problem, which I -think- has something to do with some stuff
in that backup.

Wherever the problem came from, I can't follow the instructions I found
while it persists, because attempting to run slonik_init_cluster causes the
program to error out with a message of 'Error: namespace "_cpc_replic"
already exists in node 1', followed by a second error of "ERROR: no admin
conninfo for node 134586768".

The program that generated the backup I'm using grabbed the public schema
and public data, storing them in separate files that were then packaged
together.  In theory, this should not have been a problem, but in practice,
I've somehow got a fair number of operators and triggers designed for the
replication schema (I know that's what they are because they've got the
schema's name on them; it's how I found them in the public schema's file in
the first place).  They don't appear to actually be in the database, as
loading a fresh db with a copy of the schema generates error output for each
of them (operators get 'operator "x" already exists', followed by two
'schema "_cpc_replic" does not exist' messages; the triggers simply get a
single copy of the 'schema does not exist' message), and because when I go
in, even though they were generated as public.* operators and triggers, I
can't drop them, getting the same 'schema does not exist' message.

I may be barking up the completely wrong tree with where I'm going; I don't
know enough about Slony to say.  Given that the schema in question doesn't
actually exist in the database, I don't appear to be able to use the DROP
SCHEMA CASCADE recommendation, and I'm not sure that would have any effect
anyway, given that they appear to be part of the public schema, which I'd be
scared to drop.

Will I have to manually edit those lines involving the replication schema
out of the schema file I'm loading from, or is there some way to remove the
old namespace from my setup under these conditions?  And if there is a way
to do so, where would I go to look it up?

Thank you for your consideration.
-- 
View this message in context: http://www.nabble.com/How-to-correct-%22Error%3A-namespace-x-already-exists-in-node-1%22-error-when-starting-slon_init_cluster--tp19435796p19435796.html
Sent from the Slony-I -- General mailing list archive at Nabble.com.