Sanity check: db error handling

Question

I'm changing some code to return a better error on a record not being found. Whil I'm in there I find the error handling to be suspect:
                } finally {
                    resultSet.close();
            } finally {
                statement.close(); // etc, throws exception on any problem
        } catch(Throwable e) {
            e.printStackTrace();
            /* wrap and rethrows exception */               
        }Catching Throwable is a red flag to me (anyone disagree) but I don't want to mess with that.
The thing that concerns me is that if there is an exception closing the DB resources, that exception will be reported to the client process as a failure even when the daa was successfully retrieved. Or that exception will mask the actual DB error. This isn't just academic becasue we are using Oracle, and I know that sometimes Oracle will throw an exception on close that amounts to "already closed" which is not an error in my book.
Does anyone have any good reason why the way it's being done shouldn't be messed with? I've never seen this approach and I want to be sure I'm not just missing something.

Answer 1

Aye, it seems reasonable to throw up any exceptions generated by the execution of your SQL statement to the caller, but maybe you just want to log and move on should you generate exceptions from your close calls. Should your pool get exhausted or your number of connections to your DB start to grow exponentially those logs would be worth looking at.
On a related note - is it reasonable to surround your base Servlet/Struts action class's "doSomeWork" method with a try/catch block that catches Exception (even though catching Exception throws up a red flag), on the position that you want to log anything that managed to get that far, and not show the user the ol' 500 code (though I think a specified error page would assist there as well). The assumption is that any "expected" Exception (if you'll permit me that oxymoron) would've been explicitly caught and handled by now, so this is something truly exceptional .
But then again, one could argue that if something that unexpected has occurred, you should make it blow up as big as possible 'cause you should have dealt with it. But then, I don't think it looks very professional to have that big blow up happen all over a user's browser.