Instructions for failing over:

If the primary machine is still up:

1.(a)  ifdown the service IP on the primary
		/etc/init.d/service_ip.sh stop
1.(c)  make sure that archive directory on the primary is NFS mounted
		on the standby.  Should be set up already by fstab.
2.(a)  ifup the service IP on the standby
		ifup --allow service eth0:1
2.(b)  modify the pg_hba.conf file to accept client connections:
		sed -i 's/#failover#//' /etc/postgresql/9.1/main/pg_hba.conf
       if the former active is up, those two connection lines in pg_hba.conf
       should be disabled.
       then run "pg_ctlcluster 9.1 main reload" to reload pg_hba.conf file
2.(c)  run pg_ctlcluster promote on the standby, promoting it to primary
		pg_ctlcluster 9.1 main promote
2.(d)  create /srv/postgresql/9.1/main/active-trigger on the standby if
		it doesn't exist already
