Mailing List Archive

On your blog-posting you wrote: "However, mildew has identified
that the segfault seems to occur only on 64 bit machines, which
is a very good hint. "

But my segfaulting machine is an IBM x345, which doesn't have
a 64-bit cpu.. Haven't had a segfault since October fifth, so the
"export MALLOC_CHECK_=2" seems to have made it more stable.. still
hoping for a crash soon to get a good backtrace..



-jf

Oops, sorry, that was my fault. Mildew made me aware of the x64 tendency
and asked if I had any other (non-x64 reports). I said no - and
obviously screwed up. So it is me to blame ;)

And: thanks for the feedback, this brings us onto the right track!

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, October 12, 2007 9:15 PM
> To: rsyslog at lists.adiscon.com
> Subject: Re: [rsyslog] rsyslog 1.19.9 officially out
>
>
> On your blog-posting you wrote: "However, mildew has identified
> that the segfault seems to occur only on 64 bit machines, which
> is a very good hint. "
>
> But my segfaulting machine is an IBM x345, which doesn't have
> a 64-bit cpu.. Haven't had a segfault since October fifth, so the
> "export MALLOC_CHECK_=2" seems to have made it more stable.. still
> hoping for a crash soon to get a good backtrace..
>
>
>
> -jf
>
> _______________________________________________
> rsyslog mailing list
> http://lists.adiscon.net/mailman/listinfo/rsyslog
>

Hi Rainer,

> Hi all,
>
> Rsyslog 1.19.9 has been officially released today. The new release
> primarily focuses on the build system, which now again comes as a single
> tarball. Also, there is some code restructuring done and some
> experimental code that shall aid in detecting a bug occurring in some
> environments. If you are happy with your current release, there is no
> need to update to this one.
>
> Many thanks to everyone who helped with creating this version.
>
> I would also deeply appreciate if those of you that experience the
> segfault could try out the new version and report an any problem (or
> absence of it) they see.

I've applied this version on all 6 servers that use rsyslog.

I create RPM's and then apply them, part of the RPM installation is to restart
rsyslog. All 32bit servers restarted fine, both x86_64 bit servers (running
hyperthreaded) gave the following response:

Oct 13 10:56:28 server kernel: Kernel logging (proc) stopped.
Oct 13 10:56:28 server kernel: Kernel log daemon terminating.
Oct 13 10:56:29 server rsyslog: rklogd shutdown succeeded
Oct 13 10:56:29 server rsyslogd: [origin software="rsyslogd"
swVersion="1.19.7" x-pid="2671"] exiting on signal 15.
Oct 13 10:57:23 server rsyslogd: [origin software="rsyslogd"
swVersion="1.19.9" x-pid="11827"][x-configInfo udpRecep
tion="No" udpPort="514" tcpReception="No" tcpPort="0"] restart
Oct 13 10:57:23 server rsyslog: rsyslogd startup succeeded
Oct 13 10:57:23 server kernel: rklogd 1.19.9, log source = /proc/kmsg started.
Oct 13 10:57:23 server kernel: rsyslogd[11109]: segfault at 000000000000000a
rip 00000038c7568678 rsp 00000000409fe0
60 error 4
Oct 13 10:57:23 server rsyslog: rklogd startup succeeded
Oct 13 10:57:23 server rsyslog: rsyslogd shutdown succeeded

As we've discussed before however, I've never actually had rsyslog crash on
these servers before, I only get this segfault when I restart (and only on the
first restart on installing the RPM, consecutive restarts don't segfault).

I've also updated the wiki.

Thanks,

Michael.

> Change Log:
> http://www.rsyslog.com/Article135.phtml
>
> Download:
> http://www.rsyslog.com/Downloads-req-getit-lid-61.phtml
>
> As always, feedback is appreciated.
>
> Rainer Gerhards
> _______________________________________________
> rsyslog mailing list
> http://lists.adiscon.net/mailman/listinfo/rsyslog
------- End of Original Message -------