On 20/10/2019 21:15, Johannes Edmeier wrote: > Am So., 20. Okt. 2019 um 16:11 Uhr schrieb Mark Thomas <ma...@apache.org>: >> >> On October 20, 2019 9:43:00 AM UTC, Johannes Edmeier >> <johannes.edme...@gmail.com> wrote: >>> Hi folks, >>> >>> I have a async response which is designed to run as long as the >>> browser is connected (using SSE). >>> After the browser disconnects and data is written to the response >>> internally a "IOException: Broken Pipe" is thrown, which itself is not >>> logged, but there is some error from the state handling logged (see >>> below) which is quite noisy. Is there somehow to prevent this from >>> happening? or is this a bug? >> >> It may be an application bug or a Tomcat bug. The Tomcat side of things has >> been cleaned up for 9.9.28. >> >> Mark >> > > Thanks for your quick response. I did test it with the 9.0.28-dev and > indeed the error isn't logged anymore. > Thanks a lot and keep up the good work!
Thanks for the feedback. It is worth pointing out (as this thread seems as relevant as any) that as soon as you see an I/O error on an async thread the expectation is that: - the async thread exists immediately (well, doesn't attempt to read from the request, write to the response, call complete() etc.) - Tomcat will call the onError() method of any attached AsyncListeners - One of those listeners will call complete() or dispatch() - If there is no call to complete() or dispatch(), Tomcat will call complete() It is that first step that usually where things start to go wrong. Mark --------------------------------------------------------------------- To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org For additional commands, e-mail: users-h...@tomcat.apache.org
