adrian,
thanks for the new version of sylkserver. i built it myself on debian
squeeze by first getting it from your repo with command
apt-get source sylkserver
build and install went fine, but when i try to start sylkserver, i get
the error below.
any idea what goes wrong? my python-sipsimple
Bogdan-Andrei Iancu writes:
> The backtrace is corrupted and the only useful info I can see from there
> is that the crash is related to libxml2 library (used for parsing the
> XML bodies).
>
> The logs shows that the crashed process is the timer (if I'm not wrong)
> - last messages show the D
Bogdan-Andrei Iancu writes:
> Is there a way to reproduce such crash ?
bogdan,
a SIP UA vendor did some presence tests when the crashes happened. i
asked them to try again today, but so far i have not seen the crashes.
one of reasons might have been malformed xml bodies in publish
requests.
be
today my presence server based on quite recent opensips source from svn
trunk crashed several times. backtrace of each crash looked liked this:
Program terminated with signal 11, Segmentation fault.
[New process 2447]
#0 0xb74ad4cb in strlen () from /lib/i686/cmov/libc.so.6
(gdb) where
#0 0xb74
Anca Vamanu writes:
> Yes, now referencing resource-list documents in a rls-services
> documents is not supported. But we should implement this as it is much
> requested. Can you please make a feature request entry with this on
> sourceforge?
anca,
i just added such a feature request and men
i have a sip client that instead of listing members of a list in
rls document, it puts there a reference to another document in the same
xcap server that contains the member info:
http://xcap.test.fi:8080/xcap-root/resource-lists/users/sip:j...@test.fi/generallist.xml/~~/resource-lists/list...@
IƱaki Baz Castillo writes:
> So such PUT will be rejected, but the existing document (containing
> external references) will remain *active*.
perhaps you can write an external script so remove the external
references when you change configuration of xcap server. changing the
config file is an ex
i have noticed a few times opensips 1.6 crashing while doing some timer
presence/related db operations. below is gdb output of such crash.
-- juha
Program terminated with signal 11, Segmentation fault.
#0 0xb76d23e3 in db_mysql_val2bind (v=0xbf96715c, binds=0x8188a18, i=1)
at val.c:274
274
Juha Heinanen writes:
> could someone please fix the schema and also it would be nice if the
> extra space could be eliminated like i suggested earlier (see below).
i increased rls_presentity content_type field to char(255) and after
that (plus after fixing ;start=), wireshark s
Juha Heinanen writes:
> i then went to see how multipart body is generated and found that its
> parts are taken from rls_presentity table and indeed start value is
> bogus already there:
>
> -- Dumping data for table `rls_presentity`
>
> INSERT INTO `rls_presentit
i then went to see how multipart body is generated and found that its
parts are taken from rls_presentity table and indeed start value is
bogus already there:
-- Dumping data for table `rls_presentity`
INSERT INTO `rls_presentity` (`id`, `rlsubs_did`, `resource_uri`, `content_type`
, `presence_st
Juha Heinanen writes:
> i turned LM_DBG to LM_INFO in rls_send_notify just before the request is
> sent out:
>
> LM_INFO("str_hdr= %.*s\n", str_hdr.len, str_hdr.s);
> rt = tmb.t_request_within
> (&met,
> &str_hdr,
&
anca,
i did some debugging related to the bogus multipart/related mime header
in rls subscribe.
i turned LM_DBG to LM_INFO in rls_send_notify just before the request is
sent out:
LM_INFO("str_hdr= %.*s\n", str_hdr.len, str_hdr.s);
rt = tmb.t_request_within
(&met,
Anca Vamanu writes:
>
> I don't think there is a '\r\n' in the middle for the start string - it
> is a dot '.'
> Here is where the string is generated:
>
> sprintf(cid, "%d.%.*s.%d", (int)time(NULL), uri_len, uri, rand());
anca,
perhaps there is a misunderstanding. i don't know if there i
Anca Vamanu writes:
> However I see that my wireshark still shows
> \r\n at the beginning of the boundary string...
anca,
don't worry about that, because it is how it should be:
Note that the encapsulation boundary must occur at the beginning of a
line, i.e., following a CRLF, and that that in
Ovidiu Sas writes:
> I don't think that the second boundary looks ok:
> Boundary: \r\n--1fZ25o5K6UJOKcK25Mw2VZNJ\r\n
>
> See the \r\n in the middle of the string.
yes, and also start param value looks wrong on this line:
Content-Type: multipart/related;type="application/rlmi+xml";start=
"<1275
Adrian Georgescu writes:
> Is also possible that being UDP, the large packet did not entirely
> traverse or was not correctly reassembled by the NAT-ed router.
adrian,
this is not the case, because the test was run on a single host and no
nat was involved.
so either wireshark does not know how
Anca Vamanu writes:
> Thank you for the report. I have just committed a fix for this. Please
> update.
anca,
thanks for the fix. now to uri is ok. the only remaining thing that
bothers me is that wireshark is not able to decode the notify send out
by opensips presence server (packet dump belo
i created s resource list jh-budd...@test.fi, which contains two uris:
t...@test.fi and test-budd...@test.fi.
then i tried to subscribe it, but opensips complained about parsing of
to header.
syslog shows that rls_handle_subscribe is called on jh-budd...@test.fi:
May 29 18:45:49 localhost /usr/s
Adrian Georgescu writes:
> mediaproxy (2.4.0) unstable; urgency=low
>
>* Added ICE negotiation capability and assert itself as a TURN relay
> candidate in the SDP, for a complete description and usage
> information
> see http://mediaproxy-ng.org/wiki/ICE
>* Fix race in
Juha Heinanen writes:
> adrian,
>
> the list does not include anything about python-application
> 1.2.2 compatibility. does it work with that version or still require
> old version?
>
> -- juha
sorry about the mixup. the problem was with openxcap, not
Bogdan-Andrei Iancu writes:
> This does mean you encountered a deadlock. By the sounds of it, you're
> likely to be using InnoDB.
bogdan,
yes, because myisam is not good for drbd. that reminds me, why were all
vachars changed to chars?
> So, you should issue "SHOW INNODB STATUS" from the m
i tried opensips 1.6 presence and once in a while i get to syslog these:
Dec 7 11:48:14 lohi /usr/sbin/pres-serv[24449]:
ERROR:db_mysql:db_mysql_do_prepared_query: driver error: Deadlock found when
trying to get lock; try restarting transaction
Dec 7 11:48:14 lohi /usr/sbin/pres-serv[24449]:
Dan Pascu writes:
> In the end it means, the proxy will have to verify everything (dialog
> identification elements, cseq, ruri, route set) to avoid fraud and also
> wait for a 200 OK, which makes it look more like a b2bua after all
and if gw is down and 200 ok never comes. the user should
Anca Vamanu writes:
> It's a good idea and I think it is useful now since only very few
> clients support Xcap auth. I will put it on the todo list
if i remember correctly, it would have been easy to implement for
initial subscribes, but more difficult for in-dialog subscribes, because
there w
Anca Vamanu writes:
> You must not set the force_active parameter - remove the line
> modparam("presence_xml", "force_active", 1) from your configuration
> file. Default value for it is 0 - meaning that the authorization state
> is computed according to the rules in xcap documents and that
26 matches
Mail list logo