I was trying out @RoundRobinPartitioning with EclipseLink. I am seeing a weird issue. I have a test which inserts mock data in Postgres 9.2 (I have two instance of Postgres 9.2 running). The replication works fine (I have set replicateWrites to true). But when I run test queries on the cluster – the same queries runs slower on one instance vs. another. When I looked at the query plan both databases generated different query plans for the exact same data/indexes etc.
Any ideas why this would happen?
Thanks!
Never mind – running Analyze and Reindex on the table solved the problem.