Uploaded image for project: 'Debezium'
  1. Debezium
  2. DBZ-1426

Possible issue with Debezium not properly shutting down PG connections during Connect rebalance


      Hello, we have a fairly large K8s Connect cluster with 100s of Debezium 10.0 Beta 3 PG Connectors running against PG 9.6 servers using decoderbufs. As we add new connectors, Connect rebalances and the rebalancing causes connectors to disconnect and try to reconnect to the PG replication slot that they were streaming to. Often during this process the Connector crashes with an error about the slot already being used. Each Connector is using a unique name and there should be no chance of overlap. This crash causes another rebalance to happen and things can churn for quite awhile. Eventually things settle down and everything connects. Is Debeizum properly handling replication slot connection shutdowns when rebalancing happens? Is there some setting in PG that needs to be changed? Is anyone else seeing similar issues related to unclean disconnects?

            Unassigned Unassigned
            trolison Taylor Rolison (Inactive)
            0 Vote for this issue
            3 Start watching this issue