[Serdev] Re: [Serusers] sems - answ_machine fails with "signal 17 received"

Ulrich Abend ullstar at iptel.org
Tue Jul 29 08:52:52 UTC 2003


Klaus,

comments inline.

Am Montag, 28. Juli 2003 20:21 schrieb Klaus Darilion:
> I have tried it with ulimit -c unlimited, but sems produces no core
> dump, sorry.
ok. I ran some more extensive tests and can now completely reproduce your 
problems on redhat 9. All other plattforms I tried seem to work fine btw, 
what makes me think it has to do with redhat's new native posix thread 
support.
You have three opportunities now, wait until I fixed the problem (might take a 
little, debugging threading problems is a big pain), fix it oin your own 
(that would be greatly appreciated ;-) or use another plattform for voicemail 
(best redhat ever: 7.3, gentoo, suse, freebsd>=5.1, ....)

> Another question: Does sems requires root privileges? When I try to
> start sems as normal user, the following error message occurs:
> (9627) ERROR: write_pid_file (ans_machine.cpp:93): could not write pid
> file: Permission denied
No, it does not require root privileges, you can even run it from within the 
source tree without problems. The error concerning the pid file is no 
problem, as long as you make shure, you have only one ans_machine running on 
one configured am_fifo.

> So, I just installed the stable 0.8.11 branch (ser + sems), and the
> problem is still the same: sometimes the sessions are not cleaned
> (session ... still running), and sometimes sems aborts after the first
> received call with "signal 17 received". The errors appears to be not so
> often as with the current CVS, but I think that's just random.
You are right it's random, see above.

Regards,

Uli.


>
> regards,
> Klaus
>
> > -----Original Message-----
> > From: Ulrich Abend [mailto:ullstar at iptel.org]
> > Sent: Monday, July 28, 2003 6:23 PM
> > To: Klaus Darilion
> > Cc: serdev at lists.iptel.org
> > Subject: Re: [Serusers] sems - answ_machine fails with
> > "signal 17 received"
> >
> >
> > Hi Klaus,
> >
> > /* moved the discussion to the serdev list */
> >
> > comments inline.
> >
> > Am Montag, 28. Juli 2003 17:16 schrieb Klaus Darilion:
> > > Have you changed something on the code yet?
> > >
> > > The newest file in CVS is ans_machine.cpp and it is
> >
> > reported as 39hours
> >
> > > old. (I use the web-cvs to download the tarball). Do I have
> >
> > used an old
> >
> > > version again?
> >
> > No, I only changed some build parameters yet, I just wanted
> > to make shure you
> > use the 39h old changes ;-) Downloading the tarball should be ok.
> >
> > Am Montag, 28. Juli 2003 17:09 schrieb Klaus Darilion:
> > > Then, I started sems again. I called into the echo module again, and
> > > sems terminated immediately with "signal 17 received". There was no
> > > segmantation fault (and I never had any), so there is no core dump.
> >
> > The signal is caught, so you will not see a segmentation
> > fault message. Have
> > you tried with ulimit -c unlimited?
> >
> > > The log of the second run of sems is am4.txt, and the
> >
> > corresponding ser
> >
> > > log is ser4.txt.
> > >
> > > The strange thing is, that the "signal 17" error occurs only after a
> > > restart of sems. Is it necessary to restart ser too, after
> >
> > I restarted
> >
> > > sems? Are there any old commands in the fifo that may cause
> >
> > problems? Is
> >
> > > it necessaryto start ser/sems in a certain sequence?
> >
> > AFAIK no, I don't restart ser for days, only sems.
> >
> > Please try the stable branches as described in the last mail.
> > I am quite busy
> > at the time and Raphael, the original developer of sems, is
> > unavailable for
> > the next two weeks. I will do what I can to resolve the
> > issues, but I cannot
> > promise anything.
> >
> > Regards,
> >
> > Uli.
> >
> > > > -----Original Message-----
> > > > From: Ulrich Abend [mailto:ullstar at iptel.org]
> > > > Sent: Monday, July 28, 2003 4:15 PM
> > > > To: Klaus Darilion
> > > > Cc: serusers at lists.iptel.org
> > > > Subject: Re: [Serusers] sems - answ_machine fails with
> > > > "signal 17 received"
> > > >
> > > >
> > > > Hello Klaus,
> > > >
> > > > I could reproduce the problem with the session cleaner, what
> > > > I did not get is
> > > > the CTRL-C segfault problem. It would be very helpful if you
> > > > could update
> > > > your cvs again and create a core dump of this if you can
> >
> > reproduce it.
> >
> > > > Enter
> > > > ulimit -c unlimited
> > > > before starting sems, this enables core dumps to be written.
> > > > If sems ends with
> > > > "signal 17 received" a file named core-XXX should be created.
> > > > Then tar the
> > > > whole anwer_machine directory and send it to
> > > > serhelp at lists.iptel.org. Thanks.
> > > >
> > > > I will address these problems as soon as I have some time...
> > > >
> > > > In the meanwhile you might want to use the stable branch,
> > > > which is known to
> > > > work quite well. You have to get the SER *and* SEMS stable branch.
> > > >
> > > > mv sip_router sip_router.unstable
> > > > cvs co -r rel_0_8_11 sip_router
> > > >
> > > > mv answer_machine answer_machine.unstable
> > > > cvs co -r ser_rel_0_8_11 answer_machine
> > > >
> > > > Regards,
> > > >
> > > > Uli.
> > > >
> > > > Am Montag, 28. Juli 2003 14:52 schrieb Klaus Darilion:
> > > > > Hello!
> > > > >
> > > > > I still have some problems with sems (i started ser and sems in
> > > > > no-daemon mode). I tried the echo mode. When I call into
> > > >
> > > > the echo module
> > > >
> > > > > the first time (Using Windows Messenger), everything works
> > > >
> > > > fine. Also
> > > >
> > > > > the second time everything works fine. But up from the
> > > >
> > > > third call, I do
> > > >
> > > > > not receive RTP any more. To me, it looks like the
> >
> > session clean-up
> >
> > > > > doesn't work correctly (refer ser1.txt and sems1.txt):
> > > > > (30881) DEBUG: run (AmSession.cpp:256): Session cleaner finished
> > > > > (30881) DEBUG: run (AmSession.cpp:225): Session cleaner
> >
> > starting its
> >
> > > > > work
> > > > > (30881) DEBUG: run (AmSession.cpp:242): session 80c3c48
> > > >
> > > > still running
> > > >
> > > > > (30881) DEBUG: run (AmSession.cpp:242): session 80caca0
> > > >
> > > > still running
> > > >
> > > > > Then I tried the same scenario a second time. I interrupted both
> > > > > applications with CTRL-C, then I started sems again,
> >
> > and then ser.
> >
> > > > > Then, when I called in again, sems stop processing with
> >
> > "signal 17
> >
> > > > > received".
> > > > > (please refer to ser2.txt and sems2.txt).
> > > > >
> > > > > I often have this problem, then I restart sems and ser, and
> > > >
> > > > sometimes,
> > > >
> > > > > after several restarts, everything works fine again.
> >
> > This problem
> >
> > > > > appears and dissapears randomly.
> > > > >
> > > > > I really love this application so I would be thankful
> >
> > for any hints.
> >
> > > > > Klaus
> > > > >
> > > > > > -----Original Message-----
> > > > > > From: Ulrich Abend [mailto:ullstar at iptel.org]
> > > > > > Sent: Sunday, July 27, 2003 11:24 AM
> > > > > > To: Klaus Darilion
> > > > > > Cc: serusers at lists.iptel.org
> > > > > > Subject: Re: [Serusers] sems - answ_machine fails with
> > > > > > "signal 17 received"
> > > > > >
> > > > > >
> > > > > > Hi Klaus,
> > > > > >
> > > > > > the files for individual users are searched like this:
> > > > > >
> > > > > > <announce_path>/<domainname>/<username>.wav
> > > > > >
> > > > > > If this file is not available
> > > > > > <announce_path>/<default_announce> is used.
> > > > > >
> > > > > > Regards,
> > > > > >
> > > > > > Uli.
> > > > > >
> > > > > > Am Sonntag, 27. Juli 2003 09:12 schrieb Klaus Darilion:
> > > > > > > Thanx, sems works fine now.
> > > > > > >
> > > > > > > Is it also possibly to use not the default message, but a
> > > > > >
> > > > > > personalized
> > > > > >
> > > > > > > message for every user? I saw this in serweb, where
> > > >
> > > > every user can
> > > >
> > > > > > > upload his personal greetings. Is this already supported in
> > > > > >
> > > > > > ans_machine?
> > > > > >
> > > > > > > regards,
> > > > > > > Klaus
> > > > > > >
> > > > > > > > -----Original Message-----
> > > > > > > > From: Ulrich Abend [mailto:ullstar at iptel.org]
> > > > > > > > Sent: Sunday, July 27, 2003 2:33 AM
> > > > > > > > To: Klaus Darilion
> > > > > > > > Cc: serusers at lists.iptel.org
> > > > > > > > Subject: Re: [Serusers] sems - answ_machine fails with
> > > > > > > > "signal 17 received"
> > > > > > > >
> > > > > > > >
> > > > > > > > Hi Klaus,
> > > > > > > >
> > > > > > > > thanks for reporting, this should be fixed now, could you
> > > > > > > > please try again
> > > > > > > > after cvs update?
> > > > > > > > Please check, that the you have the glib2-devel rpm
> > > > > > > > installed. Voicemail works
> > > > > > > > for me on a fully patched rh9.
> > > > > > > >
> > > > > > > > Regards,
> > > > > > > >
> > > > > > > > Uli.
> > > > > > > >
> > > > > > > > Am Freitag, 25. Juli 2003 19:49 schrieb Klaus Darilion:
> > > > > > > > > Hi!
> > > > > > > > >
> > > > > > > > > I'm using todays CVS for ser and sems on redhat 9. As
> > > > > > > >
> > > > > > > > database module, I
> > > > > > > >
> > > > > > > > > use mysql. When ser routes a call to sems, sems
> >
> > fails with
> >
> > > > > > > > "signal 17
> > > > > > > >
> > > > > > > > > received".
> > > > > > > > >
> > > > > > > > > The log of ans_machine and ser are following,
> >
> > attached are
> >
> > > > > > > > the config
> > > > > > > >
> > > > > > > > > files. both fifo are read/write for everybody.
> > > > > > > > >
> > > > > > > > > The CVS version of sems didn't compiled and stoped with
> > > > > >
> > > > > > an error in
> > > > > >
> > > > > > > > > ans_machine.cpp: 'pair' was not declared in this scope
> > > > > > > > > I solved this error by adding
> > > > > > > > > 	#include <vector.h>
> > > > > > > > > in ans_machine.cpp. Could this be the cause of
> >
> > the error?
> >
> > > > > > > > > regards and thank you for any hints,
> > > > > > > > > Klaus
> > > > > > > > >
> > > > > > > > > ----------------
> > > > > > > > > ans_machine log
> > > >
> > > > [removed rest of debug output]




More information about the Serdev mailing list