Re: Cache lookup failed for relation, when trying to DROP

On 10/15/2004 4:20 AM, Mark Gibson wrote:

> Andrew Sullivan wrote:
>> On Wed, Oct 06, 2004 at 05:25:58PM +0100, Mark Gibson wrote:
>>
>>>I had to remove Slony's schema manually as I was having problems
>>>with it. I was in the process of removing all Slony related stuff,
>>>and all my slave tables when this problem occurred, and was going to
>>>start again from scratch.
>>
>> Did your problem happen on a replica, or on the origin?  There's a
>> current dirty, evil hack in Slony that does extremely naughty things
>> in the catalogues on the replicas.  This is slated to go away in the
>> future, but at the moment it's possible to trip over it if you don't
>> use Slony's own admin tools.
>
> Yes it was on the slave. After a bit more playing with Slony, I've
> discovered the cause. I'd created rules on a slave table before
> subscribing it to the master, Slony was disabling the rule from
> within pg_catalog, so when I manually removed Slony I had some
> rogue rules floating around. PostgreSQL didn't know it needed to
> drop the rules but it was being restricted from dropping the table
> by unknown deps in pg_depend.
>

Yes, this is the ugly bit of catalog scrbbling Slony-I 1.0 does. We have
ideas to clean this up in 1.1.


Jan

--
#======================================================================#
# It's easier to get forgiveness for being wrong than for being right. #
# Let's break this rule - forgive me.                                  #
#================================================== JanWieck@Yahoo.com #