Thanks for the feedback. I will probably release a new version today. It
has an important fix, which hopefully solves this issue. The bad thing
is that I can not reproduce the problem in my lab, so I am basically
back to reviewing code and listening to your feedback ;) I have one more
area (in the same class) under suspicion. But maybe I do not change that
before trying out the current code change.
As a side-note, the *actual* root cause was a too-complex internal API,
which lead to wrong calling sequences in some parts of the code. I have
now re-structured the API and revisited all places where it was called.
There is another similar API and this is what I am currently reviewing.
I am not sure I like to change that API without real need, because it is
used a lot and any such change of course has new bug potential.
Rainer
> -----Original Message-----
> From: rsyslog-bounces at lists.adiscon.com [mailto:rsyslog-
> bounces at lists.adiscon.com] On Behalf Of Jan-Frode Myklebust
> Sent: Friday, September 07, 2007 8:51 AM
> To: rsyslog at lists.adiscon.com
> Subject: Re: [rsyslog] v1.19.1 is crashing
>
> On 2007-09-05, Jan-Frode Myklebust <janfrode at tanso.net> wrote:
> > On 2007-09-04, Rainer Gerhards <rgerhards at hq.adiscon.com> wrote:
> >>
> >> I'd still appreciate if you could apply 1.19.4 and tell me the
> outcome.
> >> I am now working on fixing the root cause. That might take a short
> >> while, as I am thinking about the best *design* to fix the issue.
> >
> > OK, thanks. I've upgraded my loghost to 1.19.4 now, will let you
> > know if it fails again.
>
> It failed again yesterday:
>
> *** glibc detected *** rsyslogd: corrupted double-linked list:
> 0xb7209028 ***
> ======= Backtrace: =========
> /lib/libc.so.6[0x4152ce3e]
> /lib/libc.so.6(cfree+0x90)[0x415305d0]
> rsyslogd(MsgDestruct+0x73)[0x8057e93]
> rsyslogd[0x804de4a]
> rsyslogd(llExecFunc+0x3f)[0x805eb0f]
> rsyslogd[0x804d8aa]
> rsyslogd[0x804d9d7]
> /lib/libpthread.so.0[0x416112db]
> /lib/libc.so.6(clone+0x5e)[0x4159414e]
>
> I have "mon" monitoring that rsyslogd is running, and restart it when
> it fails. "mon" restarted rsyslogd twice (Thu Sep 6 20:38, and Fri
> Sep 7 03:27), but I can't find any backtrace from the second crash..
>
>
> -jf
>
> _______________________________________________
> rsyslog mailing list
> http://lists.adiscon.net/mailman/listinfo/rsyslog