I am using JBoss AS 4.2.3 in combination with Postgres.
My AppServers are clustered and use a ha-local-tx-datasource to map against two PG databases (primary A and secondary B).
I am using standard EJB3 with an EJB3 Cache.
Now a scenario is needed where when the primary db fails, the system will fail-over to the other one.
This works when A fails over to B.
Now if I bring B back in, and then let A fail, the system will not switch back to A.
1) Am I missing something?
2) Will this even work?
3) I currently have the idle-timeout-minutes set to 1 and if I set it to 0 it seems to mean never. Though half a minute (1/2 idle time, what the doc claims) is still a lot of time in a real-time environment.
The doc also states that the 1/2 comes from the IdleRemover value, but where can I set it, if at all?
4) Would it be theoretically possible to be able to dynamically configure the connection-url in some way?
From what I read something like this would only be available in JBoss 5. Is this correct?