Thanks for the advice Eric. It seems from DERBY-151 that calling Thread.interrupted() before calling commit() is a reasonable workaround until the problem is fixed. Is this your assessment?
Might also be desirable to save the result and re-interrupt afterwards if interrupt status must be preserved (which it generally should). Though perhaps the same problem might be manifested in other parts of the driver. Thanks, Jim > -----Original Message----- > From: Eric Floehr [mailto:[EMAIL PROTECTED] > Sent: Wednesday, February 13, 2008 11:39 AM > To: Derby Discussion > Subject: RE: NullPointerException in Derby driver (from at least 10.2.2.0) > > Jim, > > That's definitely what we are seeing, and our only option was to restart > the app as well. You should really check the interrupt, because it > wasn't so much that InterruptedException was thrown at your code in the > thread for us. What was happening is the interrupt flag was getting set > down in the network layer, the interrupt was successfully handled there, > but the interrupt flag was not getting cleared. So then, next time > there was a commit on the database (in a totally separate part of the > code) on that thread, we had the problem. > > -Eric > > > -----Original Message----- > > From: Jim Newsham [mailto:[EMAIL PROTECTED] > > Sent: Wednesday, February 13, 2008 2:30 PM > > To: 'Derby Discussion' > > Subject: RE: NullPointerException in Derby driver (from at least > > 10.2.2.0) > > > > > > I have been seeing this same problem (very) intermittently as well... > > just > > have been way too busy with other things at the time to look into it. > > When > > it occurs, I generally see a NullPointerException on commit, and > > subsequent > > usage of the (statement? connection? data source? not sure exactly) > > gives > > "No current connection" errors. Our application needs to be restarted > > to > > recover, so I think the data source is no longer usable. > > > > This is with embedded 10.3.2.1. Not sure if the thread is being > > interrupted > > in this case. I think not, but could be wrong. > > > > Jim > > > > > -----Original Message----- > > > From: Daniel Noll [mailto:[EMAIL PROTECTED] > > > Sent: Tuesday, February 05, 2008 6:17 PM > > > To: Derby Discussion > > > Subject: Re: NullPointerException in Derby driver (from at least > > 10.2.2.0) > > > > > > On Wednesday 06 February 2008 04:00:26 Eric Floehr wrote: > > > > All, > > > > I'm experiencing a NullPointerException (somewhat random) > > apparently > > > > within the Derby JDBC driver (stack trace below), and am wondering > > if > > > > anyone has any suggestions on finding the root cause. I am using > > Derby > > > > 10.3.2.1 (though the problem also occurred on 10.2.2.0) with an > > IBATIS > > > > front-end (version 2.3.1.700, though it also occurred with 2.2), > > using > > > > Derby in single-user mode (direct not via server). > > > > > > Interesting, I found exactly the same problem. I'm not sure whether > > I > > > posted > > > about it to the list or whether I stayed quiet. > > > > > > Do you have another warning earlier in the log about the connection > > having > > > been closed? The few times I've seen this, the connection has > > "closed > > > itself" (despite being embedded, and thus having no good reason > > to...) and > > > then I've seen these errors for all operations directly afterwards. > > > > > > Daniel > > > > >
