On 6/25/2007 5:01 PM, Craig James wrote:
> Oops!  Sorry about that last message, I sent from the wrong email, then resent the reply rather than this!  Mea culpa.  Here's what I should have sent...
> 
> Andrew Hammond wrote:
>> On 6/25/07, Craig James <craig_james at emolecules.com> wrote:
>>> I'm having trouble dropping and re-creating a node.  I'm trying to 
>>> develop some scripts that will handle a failover situation and the 
>>> subsequent cleanup.  When dropping a node, here's what I'm doing:
>>>
>>>     slonik <<_EOF_
>>>     cluster name = global_cluster;
>>>     node 1 admin conninfo = 'dbname=global host=host1 user=postgres';
>>>     node 2 admin conninfo = 'dbname=global host=host2 user=postgres';
>>>     uninstall node (id = 2);
>>>     _EOF_
>>>
>>>     ... drop the slave database
>>>     ... recreate the slave database schema
>> 
>> Are you waiting for this event to propagate? Because it looks like
>> you're just running this second script immediately.
> 
> Yes, I am waiting.  The "script" I showed is actually four separate steps: Drop the node, destroy the database, recreate the schema, and recreate the node.  Just to be sure, I ran the first step, waited, and watched the slon(1) daemon's log.  When it showed, "cannot get sl_local_node_id - ERROR:  schema "_global_cluster" does not exist", I figured the event had propagated.

Doesn't matter. UNINSTALL NODE only cleans up the schema of node 2 in 
this case. It does not implicitly do a DROP NODE.


Jan

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