On 9/11/2004 10:21 PM, elein wrote:
> I am replicating one set of tables between a 74 and an 80b2 installation.
> 
> So, I screwed up my pg (80) installation and needed to bounce it.
> It would not shutdown with pg_ctl stop, probably because of
> the slon processes. (There were no open psql sessions, etc.)

Yupp. "pg_ctl stop" initiates a "smart" shutdown which will wait until 
all existing connections terminated by themself. It will not cancel any 
queries or force clients to disconnect.

> 
> I kill -TERM'd the slon processes for that (80) installation and still
> could not shut down postmaster with pg_ctl. So I kill -TERM'd the
> slon processes for the 2nd installation (74) and then the 80 postmaster
> died *without* a pg_ctl stop.  The 74 installation is alive and
> assumed well.

The initiated shutdown was still in progress, so when the last 
connection terminated, postmaster was able to proceed.

In case you want to get rid of that node entirely, you need to execute a 
"drop node" with another existing one as event node, plus an "uninstall 
node" against the one just dropped.


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 #