Done. Lmk if any questions. It's tuned to run for 20 seconds-ish with a fast producer and slow consumer and a separate advisory consumer running at full speed, and reproduces the stackoverflow but not the oom.
Note it calls System.setOut & setErr in order to detect exceptions that occur on the daemon threads, so could interfere with other tests in a suite -----Original Message----- From: Gary Tully [mailto:gary.tu...@gmail.com] Sent: Wednesday, June 02, 2010 6:14 AM To: dev@activemq.apache.org Subject: Re: test case for advisory messages can you attach it to the jira: https://issues.apache.org/activemq/secure/AttachFile!default.jspa?id=315 27 On 2 June 2010 03:11, Nathan Bryant <nathan.bry...@linkshare.com> wrote: > Oops! > > > > I guess it got stripped out. Anybody want? > > > > From: Nathan Bryant [mailto:nathan.bry...@linkshare.com] > Sent: Tuesday, June 01, 2010 10:08 PM > To: dev@activemq.apache.org > Subject: test case for advisory messages > > > > Hi, > > > > Attached, in the hope that somebody will find it useful, is a unified > test case for some of the slow consumer / advisory recursion issues, > ala AMQ-2533 > > > > Right now, this test is very good at reproducing the advisory stack > overflow issues that were present in 5.3.0. It's not very good at > triggering OOM issues but if it were tuned to run for long enough, that > might change. Several tuning constants are present at the top of the > test. > > > > (This is a little long at 500+ lines but is completely self contained. > It's a trimmed and repackaged version of my app code for now - possible > to trim it a little more) > > > > Nathan Bryant > > -- http://blog.garytully.com Open Source Integration http://fusesource.com
