On 04/20/2011 10:26 AM, rudi wrote:
> Database cluster state: in archive recovery
>
> and doesn't accept connections:
>
> 2011-04-20 10:17:00 CEST:[local]:u@postgres:[13099] FATAL: the database
> system is starting up
>
> even if is a hot standby.
My fault: it seems I forgot to stop the backup in my previous attempt. I
tried this procedure and now the former master acts as a hot standby as
expected:
On the former master:
a) stop
b) cleanup pg_xlog
c) set archive_mode = off and hot_standby = on in postgresql.conf
On the former slave:
a) set archive_mode = on in postgresql.conf
b) psql -c "SELECT pg_start_backup('switchover base backup')"
c) rsync -av $PGDATA --delete --exclude server.crt --exclude server.key
--exclude recovery.* --exclude postmaster.pid --exclude pg_xlog
--exclude archive $SLAVE:$PGDATA
d) psql -c "SELECT pg_stop_backup()"
Finally I started the former master.
--
rd
Peel your own image from the mirror.
Sit. Feast on your life.