logging-log4cxx-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From p...@mu.org
Subject Re: Race condition with FileAppender.
Date Sat, 30 Jun 2007 04:37:42 GMT
If you just build the small program that I sent and link it to the subversion log4cxx, then
check out the log files it creates.  They're full of junk along with valid lines too.

It does it everytime.  I'm working on a threaded project that uses 100's
of threads, and it was always working great with log4cxx, then I
upgraded to subversion source, and it started leaving garbage in the
logs.

Tonight, I finally decided to go back and reinstall the standard
log4cxx-9 libs, and link with them.  It completely fixed my problem.

Have you tried to run that program I sent you and link it against the
newest subversion log4cxx?

Please try and and let me know what you find in the log files.

That's the best example you could have of the problem.

Thanks,

Pete

On Fri, Jun 29, 2007 at 11:26:37PM -0500, Curt Arnold wrote:
> 
> On Jun 29, 2007, at 10:53 PM, pete@mu.org wrote:
> 
> >Nevermind.  This is a known bug.
> >
> >http://issues.apache.org/jira/browse/LOGCXX-129
> >
> >I should of just stayed with 97 which was working fine.
> >
> >Thanks,
> >
> >Pete
> >
> 
> 
> LOGCXX-129 only reported problems with AsyncAppender which was badly  
> flawed in log4j 1.2.13 and earlier and deadlock reports in log4j  
> started getting more frequent.  The log4cxx AsyncAppender was pretty  
> much a straight port of the early log4j equivalent and would be  
> expected to have the same problems under stress.   The new log4j  
> AsyncAppender should be ported over to log4cxx and that is one of the  
> remaining issues on a push to release.
> 
> However, your test case does not appear to involve AsyncAppender and  
> the anticipated resolution for LOGCXX-129 would have no impact.  If  
> you have any stack traces or other information to help localize the  
> problem, they would be appreciated.  Otherwise, it will be my top  
> issue to investigate when I shift to log4cxx development tomorrow.
> 

Mime
View raw message