Thomas,
On HPUX, the command to dump the stack trace for all threads is > kill -s 3 pid where pid is the process id your java process. I have tried this on an actual HPUX running JDK 1.3.1. HTH, Ceki At 09:38 21.05.2002 +0200, Thomas Porschberg wrote: >- We do not use multiple threads. We have simple one process. >- I can try to write a small program, that provoke the error > and then I have to hope that the error appears. > >Thanks for the hint with the "key combination", I have to >look for it. > > >On Tue, May 21, 2002 at 09:15:40AM +0200, Ceki Gülcü wrote: > > > > Is it possible that introducing the > > Category.getInstance(this.getClass()) statement changed the execution > > time of the thread and made a concurrency bug to become apparent? > > > > Most JVMs have a "secret" key combination that dumps the stacks of all > > threads on the console. Such dumps are very valuable to identify the > > reasons for infinite loops. I do not know the key combination for HPUX > > but all the JVMs I have worked with so far had such a key combination > > although you might have to dig a little. > > > > Can you isolate or reproduce the problem? A small test case > > reproducing the bug would be invaluable. > > > > Regards, Ceki > > > > At 07:37 21.05.2002 +0200, you wrote: > > > > >We use log4j, version API V1.1.3, on HPUX with VM > > > > > >Java(TM) 2 Runtime Environment, Standard Edition (build > > >jinteg:02/14/01-23:51) > > >HotSpot VM (build 1.0.1fcs jinteg:02/15/01-02:43 PA2.0, mixed mode) > > > > > >and encounter the following problem: > > > > > >Under very rare circumstances our application hangs and CPU load > > >increases. It looks like a loop. > > > > > >This happens in this piece of code: > > > > > >. > > >. > > >log = Category.getInstance(this.getClass()); > > >debug = log.isDebugEnabled(); > > >. > > >. > > > > > >Any hint how to solve or at least hunt the problem ? -- Ceki -- To unsubscribe, e-mail: <mailto:[EMAIL PROTECTED]> For additional commands, e-mail: <mailto:[EMAIL PROTECTED]>
