In PG9.0-b4 I have successfully set up a primary-secondary pair with streaming replication, with the secondary in hot_standby mode.
I have an application on the standby (and primary) which has constant connections open to the database. Occasionally it refreshes its cache of data from the hot_standby secondary server. I am intermittently getting and error when updating the application's cache. When I check the PostgreSQL logs with the same timestamp, I see the following messages:
FATAL: terminating connection due to conflict with recovery DETAIL: User query michg have needed to see row versions that must be removed. HINT: In a moment you should be able to reconnect to the database and repeat your command.
Now, this message seems to explain the problem, but I couldn't find anything in the streaming replication documentation indicating that this sort of thing might occur. Is there any way to avoid it? Any further information about this specific problem?