Hi Aki, It makes sense. The httpExchange has bunch of onXXXX method to take care of it. We should try to close the input stream when the exception is thrown.
I just filled a JIRA[1] for it. [1]https://issues.apache.org/jira/browse/CAMEL-5621 -- Willem Jiang Red Hat, Inc. FuseSource is now part of Red Hat Web: http://www.fusesource.com (http://www.fusesource.com/) Blog: http://willemjiang.blogspot.com (http://willemjiang.blogspot.com/) (English) http://jnn.javaeye.com (http://jnn.javaeye.com/) (Chinese) Twitter: willemjiang Weibo: willemjiang On Monday, September 17, 2012 at 5:55 PM, Aki Yoshida wrote: > I think there is an issue in JettyHttpProducer. > > public boolean process(Exchange exchange, final AsyncCallback callback) { > JettyContentExchange httpExchange = null; > try { > httpExchange = createHttpExchange(exchange, callback); > doSendExchange(client, httpExchange); > } catch (Exception e) { > // error occurred before we had a chance to go async > // so set exception and invoke callback true > exchange.setException(e); > callback.done(true); > return true; > } > > // we should continue processing this asynchronously > return false; > } > > createHttpExchange(exchange, callback) may open an input stream and > set it to httpExchange. When something goes wrong in doSendExchange, > this input stream may not be closed. So, I put a finally block there > which checks if the input stream in available in httpExchange and if > so, closing it. That released the file. Or something of this kind > could go into doSendExchange. > > regards, aki > > 2012/9/17 Kramer, Ulrich <u.kra...@sap.com (mailto:u.kra...@sap.com)>: > > Hello! > > > > in Apache Camel 2.10.1 the following route leaves the files in the source > > directory opened, if the destination is not reachable. Every second the > > number of opened files (lsof -p <pid>) is increased by 2. > > > > from("file:/tmp/inbox").to(""jetty:http://localhost:12345/xxx";) > > > > But the route > > > > from("file:/tmp/inbox").throwException(new Exception("Test")) > > > > works fine. Is this a bug?
