(Apologies this is spotted in kettle but Hans tells me if it occurs there, it will happen in hop)
As i'm heading towards moving to hop, i wondered if this had been addressed?
Anyway - If you use the postgresql bulk loader, the connection doesnt get released until the hop process exits. so if you do a bulk load, run some other pipelines, etc, other jobs, the connection lingers until hop finally exits. Lets hope no one does a bulk load in a never ending pipeline!