Replication/backup defaults

Cycling back to this topic again, but this time at the beginning of a CF.

Here's an actual patch to change:

wal_level=replica
max_wal_senders=10
max_replication_slots=20

Based on feedback from last year (
https://www.postgresql.org/message-id/CABUevEwfV7zDutescm2PHGvsJdYA0RWHFMTRGhwrJPGgSbzZDQ%40mail.gmail.com
):

There were requests for benchmarks of performance difference. Tomas has
promised to run a couple of benchmarks on his standard benchmarking setups
to give numbers on that. Thanks Tomas, please pipe in with your results
when you have them!

Security considerations about pg_hba.conf -- I avoided those by not
actually changing pg_hba.conf. Since pg_hba.conf can be changed on a reload
instead of a restart it's a lot easier to deal with. I still think changing
it to allow "postgres" the same type of connections as it does for regular
users would not be a security problem, but again thanks to it only needing
a reload it's not as big an issue.

There was the idea to have multiple sets of defaults to choose from at
initdb time. I don't see a problem having that, but it's been another year
and nobody built it. I don't think not having that is an excuse for the
current defaults. And implementing something like that is in no way
hindered by
changing the current defaults.

We were too close to beta1 -- this is why I'm sending it earlier this time
:) (Even though I intended to do it already back in September, better now
than even later)

Finally, there's the argument that we're already shaking up a number of
other things with version 10, so this is a good time to do this one as well.

--
Magnus Hagander
Me: http://www.hagander.net/
Work: http://www.redpill-linpro.com/