I'm starting to play with PostgreSQL 9.1, thank you all for this nice
and sweet piece of software.
I've two hosts in my cluster:
a) postgresql master
b) postgresql standby
I've created two tables on master:
create table test_logged (id serial, nome text);
create unlogged table test_unlogged (id serial, nome text);
Both tables appears on standby too but on standby following query:
select * from test_unlogged;
gives me following message:
ERROR: cannot access temporary or unlogged relations during recovery
I understand that unlogged table are not replicated, but I expected:
1) not see defined unlogged tables on standby
OR
2) see them void on standby and use them to store different set of
records for each standby (like web sessions) those need not to be
replicated in the cluster.
Robe on #postgresql suggest me to run another postgresql instance on
each custer host node to store local volatile data (like web app sessions).
Is it this the best option actually?
Thank you in advance, \ferz
PS: I've written some simply tests and I've seen that inserts on
unlogged tables are 10 times faster.