On Mon, Jun 10, 2013 at 11:02 AM, Niels Kristian Schjødt <
[email protected]> wrote:
> Okay, cool
>
> You mean that I should do the following right?:
>
> 1. Stop slave server
>
At this point, you don't have a slave server. Not a usable one, anyway.
If you used to have a hot-standby server, it is now simply a historical
reporting server. If you have no need/use for such a reporting server,
then yes you should stop it, to avoid confusion.
> 2. set archive_command = 'true' in postgresql.conf on the master server
> 3. restart master server
>
You can simply do a reload rather than a full restart.
> 4. run psql -c "SELECT pg_start_backup('label', true)" on master
>
No, you shouldn't do that yet without first having correctly functioning
archiving back in place. After setting archive_command=true and reloading
the server, you have to wait a while for the "bad" WAL files to get
pseudo-archived and cleared from the system. Once that has happened, you
can then return archive_command to its previous setting, and again
reload/restart the server. Only at that point should you begin taking the
new backup. In other words, steps 7 and 8 have to be moved up to before
step 4.
> 5. run rsync -av --exclude postmaster.pid --exclude pg_xlog
> /var/lib/postgresql/9.2/main/
> [email protected]:/var/lib/postgresql/9.2/main/" on
> master server
> 6. run psql -c "SELECT pg_stop_backup();" on master server
> 7. change archive_command back on master
> 8. restart master
> 9. start slave
>
> Just to confirm the approach :-)
>
Cheers,
Jeff
