?hel kenal p?eval, N, 2006-04-13 kell 09:26, kirjutas Aaron Randall:
> Hi All!
> 
> I have a question with regards to promoting a slave.  In my current setup, I
> have two nodes, the Master and the Slave.  Replication works perfectly, however
> I need to plan for outages of the Master.  In testing, if I kill the master
> server, I need a way (manual intervention is fine) to get the slave up and
> running so you can access the database.  Currently if I kill the Master and
> stop the Slon processes on the slave, I can read from the database, but can
> still not write to it.  I found the script on this page
> (http://www.linuxjournal.com/article/7834 under section "promote.sh") but I
> can't seem to get it to work.  Any ideas?

As you can't do a graceful switchover after master is gone, then - in a
two-node setup - the simplest thing would be just removing slony from
slave.

select _clusterschema.uninstallnode();
drop schema _clusterschema cascade;

You will have to set up the full replication after fixing the old master
anyhow.

-----------
Hannu