basti wrote:
> is it possible to have WAL Replication and Point-in-Time Recovery like
> follows:
>
> DB-Master -- (WAL to Slave) --> DB-Slave
> |
> |-------> (PITR to an other Server)
>
> Thanks for any help!
Sure. You can use something like the UNIX command "tee" in "archive_command"
to create a second copy of the WAL archive.
You could also use the same archived WALs for both purposes and
from several machines by putting them on a network file system.
> p.s. is there a tutorial somewhere how describe the steps todo when the
> master is crashed?
You will have to promote the standby, make sure future client
requests will go there (a connection pooler or a virtual IP address
can help here) and build a new standby.
Yours,
Laurenz Albe