I have discovered that PostgreSQL, Camel, and connection pools are a toxic combination when the schema search path isn't set on a per-user basis (which I can not do).
Part of the problem is that the Postgres JDBC doesn't want to add schema search path to the url. As a work-around, the jdbc/sql components could skip the first empty "set search path" result set. Perhaps there is a way to interact with the connection pool more directly. Currently it's just a black box from my perspective. My solution? Get rid of the connection pool for now. https://issues.apache.org/jira/browse/CAMEL-7904 ~cg