Alex Mueller

@jimmybow the [Oracle Docs](https://docs.oracle.com/en/database/oracle/oracle-database/19/netrf/oracle-net-listener-parameters-in-listener-ora.html#GUID-205F23E2-9DDA-4815-A60A-31B94E1F8787) say that rate limiting takes place per second. I assume that your test case opens too many connections within a second, so you reach the database-side...

Apologies @jimmybow, you're right. Though the code seemed correct at first sight, I can confirm the idle connection using Postgres: @hansva: While `TableInputMeta.getFields()` fails, it properly disconnects from the database...

It seems to me that the exception in`TableInputMeta.getFields()` does not escalate to the `TableInput`.

This could indeed be related. When you run the failed pipeline again, it does close the idle connection.