to figure out what data will be returned by the database (column names and types, basically), KNIME can execute the query with a
limit 1 suffix after configuration. This is so that the following nodes know what to expect, so that you can, for example, configure a following column filter node before actually executing the Database Reader. Disabling the option requires you to execute it before configuring other nodes that depend on its output. Does that make sense?
A KNIME node has two important functions: configure and execute. In the configure step the node calculates what it will output and in the execute it will actually produce the output. Configure is called when the node receives or loses input connections or when the settings change. Subsequently the following nodes are also configured, i.e. notified about possible changes in the input. As mentioned above, this makes it easier to configure nodes without actual data being present.