Rainer - Ok. Here we go. This is with the -O2 -g CFLAGS added as well as --enable-valgrind.
http://pastebin.com/eZRKGFnk -- James -----Original Message----- From: [email protected] [mailto:[email protected]] On Behalf Of Rainer Gerhards Sent: Monday, September 23, 2013 1:54 AM To: rsyslog-users Subject: Re: [rsyslog] Local Variables Usage Question On Sun, Sep 22, 2013 at 12:41 PM, Boylan, James <[email protected]>wrote: > I can definitely do that. Is there any specific options that you know > of that would disable those symbols? probably ADDing -g is what we need in your env: http://gcc.gnu.org/onlinedocs/gcc/Debugging-Options.html While we are at it, -O0 may also useful, as it often points more precisely to the code in question. But that's not an option to use in production, as disabling optimization can considerably slow down execution. > Also does --enable-valgrind alter the overall performance of Rsyslog? No. All it does is prevent modules from being unloaded, so that the symbols remain in place. For what we are looking at, it should not even be necessary (but could in extreme cases, e.g. if the problem were routed with closing TCP connections). Usually, it just ensures the memleak report at shutdown remains readable. > I'm just trying to see if I'll need to take that option out before the > rpms can go live. > > I still wouldn't let it in as a long-term option to use. Rainer > Otherwise I should be able to enable that. I'll aim to do so later > today when I'm back at my system. > > -- James > > ----- Reply message ----- > From: "Rainer Gerhards" <[email protected]> > To: "rsyslog-users" <[email protected]> > Subject: [rsyslog] Local Variables Usage Question > Date: Sun, Sep 22, 2013 3:39 am > > > > On Sat, Sep 21, 2013 at 2:50 PM, Boylan, James > <[email protected] > >wrote: > > > I've grabbed the output. It is here: > > > > http://pastebin.com/yfkVk5rk -- Search for ==8348== and you go right > > to the valgrind output. > > > > Very well, that helps. Unfortunately, symbols seem to have been > stripped from the binary, so I don't get the full stack trace that I > was interested in. Could you review your build process? > > It may help a bit if you add --enable-valgrind to the ./configure options. > > > > > > I did notice that when running in valgrind rsyslog doesn't crash at > > the point it usually does. It does show the output for valgrind in > > the same spot it would have crashed though. I'm assuming it doesn't > > crash because valgrind is catching the fault. > > > > > usually not, but what happens is that under valgrind the memory layout > is of course a little bit different. So it looks like the misadressing > hit a less important spot. But it may also have something to do with > valgrind instrumentation itself. Anyway, the report is what is really > interesting. > As soon as a violation occurs, something is wrong and that's really > all we need to know :-) > > Thanks again. I'll try to work with what I have, but a full stack > trace would be even better :-) > > Rainer > > > -- James > > ________________________________________ > > From: [email protected] [ > [email protected]] > > On Behalf Of Rainer Gerhards [[email protected]] > > Sent: Saturday, September 21, 2013 4:40 AM > > To: rsyslog-users > > Subject: Re: [rsyslog] Local Variables Usage Question > > > > On Sat, Sep 21, 2013 at 2:42 AM, Boylan, James > > <[email protected] > > >wrote: > > > > > Here is the pastebin for the full log that leads up to the > > > segmentation fault. This fault only happens when the variables are > > > in the Ruleset > that > > > is processing the incoming messages. > > > > > > http://pastebin.com/9ynNu6vV > > > > > > > Thx for that. Would it be possible that you also run it under > > valgrind control? That usually helps a lot. All you need to do is to > > use > > > > $ valgrind --log-fd=1 /path/to/rsyslogd ...your options... -nd > > > logfile > > > > With that, valgrind usually points right into the line that causes > > the abort. As you need to run rsyslog interactively, you need to > > stop the service first. > > > > If you could get me that log, troubleshooting most probably would be > > way easier. > > > > Thx, > > Rainer > > _______________________________________________ > > rsyslog mailing list > > http://lists.adiscon.net/mailman/listinfo/rsyslog > > http://www.rsyslog.com/professional-services/ > > What's up with rsyslog? Follow https://twitter.com/rgerhards NOTE > > WELL: This is a PUBLIC mailing list, posts are ARCHIVED by a myriad > > of sites beyond our control. PLEASE UNSUBSCRIBE and DO NOT POST if > > you DON'T LIKE THAT. > > _______________________________________________ > > rsyslog mailing list > > http://lists.adiscon.net/mailman/listinfo/rsyslog > > http://www.rsyslog.com/professional-services/ > > What's up with rsyslog? Follow https://twitter.com/rgerhards NOTE > > WELL: This is a PUBLIC mailing list, posts are ARCHIVED by a myriad > > of sites beyond our control. PLEASE UNSUBSCRIBE and DO NOT POST if > > you DON'T LIKE THAT. > > > _______________________________________________ > rsyslog mailing list > http://lists.adiscon.net/mailman/listinfo/rsyslog > http://www.rsyslog.com/professional-services/ > What's up with rsyslog? Follow https://twitter.com/rgerhards NOTE > WELL: This is a PUBLIC mailing list, posts are ARCHIVED by a myriad of > sites beyond our control. PLEASE UNSUBSCRIBE and DO NOT POST if you > DON'T LIKE THAT. > _______________________________________________ > rsyslog mailing list > http://lists.adiscon.net/mailman/listinfo/rsyslog > http://www.rsyslog.com/professional-services/ > What's up with rsyslog? Follow https://twitter.com/rgerhards NOTE > WELL: This is a PUBLIC mailing list, posts are ARCHIVED by a myriad of > sites beyond our control. PLEASE UNSUBSCRIBE and DO NOT POST if you > DON'T LIKE THAT. > _______________________________________________ rsyslog mailing list http://lists.adiscon.net/mailman/listinfo/rsyslog http://www.rsyslog.com/professional-services/ What's up with rsyslog? Follow https://twitter.com/rgerhards NOTE WELL: This is a PUBLIC mailing list, posts are ARCHIVED by a myriad of sites beyond our control. PLEASE UNSUBSCRIBE and DO NOT POST if you DON'T LIKE THAT. _______________________________________________ rsyslog mailing list http://lists.adiscon.net/mailman/listinfo/rsyslog http://www.rsyslog.com/professional-services/ What's up with rsyslog? Follow https://twitter.com/rgerhards NOTE WELL: This is a PUBLIC mailing list, posts are ARCHIVED by a myriad of sites beyond our control. PLEASE UNSUBSCRIBE and DO NOT POST if you DON'T LIKE THAT.
