In the problem situation, when $du is null after the lookup, and only y2
rings, I've seen 2 different things:
1. the INVITE to y3 is actually sent to Kamailio itself sometimes:
when Kamailio receives this, $ru of the INVITE is correctly set to the
contact of y3. Why is it sent to Kamailio itself and not y3's $ru?
-- I've seen this many times during my tests but don't know why I can't
reproduce this now. The syslog for this is in attached
failed_group_call.txt.
You can search for "route: INVITE" to see the INVITEs received by
Kamailio. This is printed out at the start of my main route.
The set up for failed_group_call.txt:
- Kamailio(a)10.82.130.244
- y1@10.82.130.135:5060
calls global.group, which has members:
- y2@10.82.130.140:19322
- y3@10.82.130.140:53444
2. the INVITE to y3 is not sent at all sometimes
The syslog for this call is in failed_group_call_notToSelf.txt
for y3@10... as you would from the other log in
1. Also I've done
tcpdump on all interfaces for this and can confirm there isn't INVITE
sent to anywhere else either, apart from to y2.
The set up for failed_group_call_notToSelf.txt:
- Kamailio(a)10.82.130.244
- y1(a)10.82.130.135
calls global.group, which has members:
- y2@10.82.130.140:39598
- y3@10.82.130.140:50104
Both log files are produced with debug set to 4.
Yufei
On 19/09/12 16:23, Klaus Darilion wrote:
On 19.09.2012 16:13, Yufei Tao wrote:
I should have made it clearer: $du is null both
before and after lookup
location in LOCATION_BRANCH when no fix_nated_register was done (thus
'received' column in location table was null). When fix_nated_register
was done, $du for each branch was null before lookup location, but set
to 'received' after lookup location as you said. But in this case
(fix_nated_register done), everything works fine.
So seems now the problem happens when: 'received' in location table is
null, causing $du not to be set. I used to think $du is only set by
record-route and thanks for clarifying this :)
Even when without fix_nated_register (thus no 'received'), the lookup
location for each branch, i.e. y2 and y3, set $ru to the 'contact' field
successfully, but $du isn't set. But Kamailio did relay to the first
branch (trunk?) and not the others. So the first branch is handled
differently? So maybe when $du is null, should set it to $ru on the
branches?
There is no need to set $du. If $du is null, $ru is used for routing.
Can you verify with ngrep/tcpdump (capture on 'any' interface) if
there is no SIP message sent to y2/y3 or the message is sent, but to
wrong destination (e.g. same target as first branch).
regards
Klaus
>
> Yufei
>
>
> On 19/09/12 11:53, Klaus Darilion wrote:
>>
>>
>> On 19.09.2012 12:24, Yufei Tao wrote:
>>> Hi Klaus
>>>
>>> Thanks for the reply!
>>>
>>> I check the $du, it is always null before and after the lookup. Is it
>>> only set when relaying to a proxy (from record-route), and not to a
>>> client?
>>
>> That's strange. For NATed clients, $du must contain the 'received'
>> URI. Otherwise they can not be contacted as $ru contains the private
>> IP address.
>>
>>> When no fix_nated_register is called, the lookup location for both
>>> clients y2 and y3 is successful from the log, when printing out $ru
>>> after lookup. But seems Kamailio only relays to one client's IP while
>>> not others. I think there must be some differences when branch
>>> route is
>>> executed first time and second time as you said. As it feels like for
>>> the first branch (trunk?) it used the 'contact' column from the
>>> location
>>> table, and for the other branches, it tries to use 'received'?
>>
>> It seems db_lookup() creates multiple branches. Is lookup() only
>> finding 1 contact in table or multiple contacts?
>>
>> regards
>> Klaus
>>
>>>
>>>
>>> Yufei
>>>
>>> On 18/09/12 18:49, Klaus Darilion wrote:
>>>> I suspect that the branch route is first executed for the NATed
>>>> client. Then the 'received' column is used as destination URI.
When
>>>> executing the branch route again, the destination URI is still the
>>>> value from the previous branch, and lookup() will not overwrite is as
>>>> 'received' is not available. Then Kamailio sends the INVITE again
to
>>>> the first client.
>>>>
>>>> You can try to set $du to Null before lookup(). ($du=null or
>>>> $du=$null, not sure what the correct syntax is).
>>>>
>>>> Another workaround is to use fix_nated_register() for every client
>>>> (the pragmatic and more secure approach).
>>>>
>>>> regards
>>>> Klaus
>>>>
>>>> On 18.09.2012 15:16, Yufei Tao wrote:
>>>>> Hi
>>>>>
>>>>> I have a strange problem on forking calls to a group of users. For
>>>>> example I have two users y2 and y3 in dbaliases, both with
>>>>> alias_username 'group'. And y2 and y3 both registers with
Kamailio
>>>>> fine.
>>>>> When I make a call to 'group' from a third client y1, what
my
>>>>> kamailio.cfg does is: do an alias_db_lookup("dbaliases"),
and
>>>>> goes to
>>>>> BRANCH_ALIASDB, where a lookup location will be done for each of the
>>>>> username resulting from lookup of dbaliases, something like this:
>>>>>
>>>>> y1--INVITE 'group'-->lookup
>>>>> dbaliase-->[BRANCH_ALIASDB]--->'y2'-->lookup
>>>>> location---¬
>>>>>
>>>>> | |-->relay
>>>>>
>>>>> ---->[BRANCH_ALIASDB]--->'y3'-->lookup
>>>>> location----
>>>>>
>>>>> The all works well as long as all clients are NAT'ed. However
when
>>>>> they
>>>>> are not NAT'ed, e.g. all on the same LAN with Kamailio, the call
>>>>> only
>>>>> goes to one of the group members, e.g. y2 only. When checking the
>>>>> log,
>>>>> it seemed to have done the dbaliases lookup fine, and each location
>>>>> lookup successfully. But Kamailio only relayed y2's IP, e.g. to
the
>>>>> client, while y3's to itself.
>>>>>
>>>>> When comparing the location table when clients are NAT'ed or not,
I
>>>>> find
>>>>> that the 'received' column is only populated when I do
>>>>> fix_nated_register. And group calls only works when
'received'
>>>>> column is
>>>>> populated. That explains why when clients are NAT'ed group calls
>>>>> work,
>>>>> as I only do fix_nated_register if nat_uac_test returns true.
>>>>>
>>>>> But if this is the only reason, if two clients register using the
>>>>> same
>>>>> username, e.g. both as y3, and when 'received' column of
location
>>>>> table
>>>>> is empty (no fix_nated_register done), I would expect a call to y3
>>>>> should also only make 1 client ring. But in fact both of them rang!
>>>>> The
>>>>> flow is like:
>>>>>
>>>>> y1--INVITE 'y3'-->lookup location for 'y3'---->
IP of 1st client
>>>>> registered as 'y3'
>>>>> |
>>>>> ---> IP of 2nd
client
>>>>> registered as 'y3'
>>>>>
>>>>> While a call to 'group' (thus dbaliases lookup took place)
under
>>>>> such
>>>>> un-NAT'ed set up made only 1 client ring.
>>>>>
>>>>> So I can make it work by always doing fix_nated_register. But
I'm
>>>>> not
>>>>> clear about these things:
>>>>> - why does a lookup of dbaliases before lookup of location make such
>>>>> difference?
>>>>> - does lookup location work differently depending on whether it is
>>>>> called from trunk or from a route called from a branch route?
>>>>>
>>>>>
>>>>> Following is relevant parts from my config file:
>>>>>
>>>>> #############################################################
>>>>> route[LOCATION]
>>>>> {
>>>>> if ( alias_db_lookup("dbaliases") )
>>>>> {
>>>>> t_on_branch("BRANCH_ALIASDB"); # in
>>>>> branch_route[BRANCH_ALIASDB],
>>>>> # call another route that
>>>>> looks up
>>>>> location,
>>>>> # if not existent, call
>>>>> drop()
>>>>>
>>>>> }
>>>>> else
>>>>> {
>>>>> xlog("L_DBG","LOCATION: not alias - go to
lookup location
>>>>> trunk\n");
>>>>> route(LOCATION_TRUNK); # normal look up location and
>>>>> sending of
>>>>> 404 etc
>>>>> }
>>>>>
>>>>> ... ...
>>>>> }
>>>>> #############################################################
>>>>> branch_route[BRANCH_ALIASDB]
>>>>> {
>>>>> xlog("L_DBG", "BRANCH_ALIASDB: $fU@$fd ->
$rU@$rd;
>>>>> Method:$rm\n");
>>>>> route(LOCATION_BRANCH);
>>>>> }
>>>>>
>>>>> route[LOCATION_BRANCH]
>>>>> {
>>>>> if (!lookup("location"))
>>>>> {
>>>>> # Drop this branch - it's going nowhere
>>>>> drop();
>>>>> }
>>>>> }
>>>>> #############################################################
>>>>> route[RELAY] {
>>>>> xlog("L_DBG","RELAY: method=$rm, callid=$ci,
cseq=$cs\n");
>>>>>
>>>>>
>>>>> #!ifdef WITH_NAT
>>>>> if (check_route_param("nat=yes")) {
>>>>> setbflag(FLB_NATB);
>>>>> }
>>>>> if (isflagset(FLT_NATS) || isbflagset(FLB_NATB)) {
>>>>> xlog("L_DBG", "RELAY: about to call
RTPPROXY\n");
>>>>> route(RTPPROXY);
>>>>> }
>>>>> #!endif
>>>>>
>>>>> /* example how to enable some additional event routes */
>>>>> if (is_method("INVITE")) {
>>>>> t_on_reply("REPLY_ONE");
>>>>> t_on_failure("FAIL_ONE");
>>>>> }
>>>>>
>>>>> if (!t_relay()) {
>>>>> sl_reply_error();
>>>>> }
>>>>>
>>>>> exit;
>>>>> }
>>>>>
>>>>> ############################################################
>>>>> route[NAT] {
>>>>> #!ifdef WITH_NAT
>>>>> xlog("L_DBG","NAT: method=$rm, callid=$ci,
cseq=$cs\n");
>>>>>
>>>>> force_rport();
>>>>> if (nat_uac_test("2")) {
>>>>> if (method=="REGISTER") {
>>>>> fix_nated_register();
>>>>> xlog("L_DBG","NAT: Just done
fix_nated_register in
>>>>> REGISTER
>>>>> message in NAT route\n");
>>>>> } else {
>>>>> xlog("L_DBG","NAT:
fix_nated_contact\n");
>>>>> fix_nated_contact();
>>>>> }
>>>>> setflag(FLT_NATS);
>>>>> }
>>>>>
>>>>> # setflag(FLT_NATS); ## -- YT: set NAT flag for all, so will
>>>>> force
>>>>> media relay
>>>>> #!endif
>>>>> return;
>>>>> }
>>>>>
>>>>> #############################################################
>>>>> And in the main route, route LOCATION and RELAY are the last two
>>>>> routes:
>>>>> route {
>>>>> ...
>>>>> route(NAT);
>>>>>
>>>>> ... ...
>>>>>
>>>>> # user location service
>>>>> route(LOCATION);
>>>>>
>>>>> route(RELAY);
>>>>> }
>>>>>
>>>>> Hope I have made it clear. Thanks very much!
>>>>>
>>>>> Yufei
>>>>> --
>>>>> Yufei Tao
>>>>> Red Embedded
>>>>>
>>>>> This E-mail and any attachments hereto are strictly confidential and
>>>>> intended solely for the addressee. If you are not the intended
>>>>> addressee please notify the sender by return and delete the message.
>>>>>
>>>>> You must not disclose, forward or copy this E-mail or attachments to
>>>>> any third party without the prior consent of the sender.
>>>>>
>>>>> Red Embedded Design, Company Number 06688253 Registered in England:
>>>>> The Waterfront, Salts Mill Rd, Saltaire, BD17 7EZ
>>>>>
>>>>> _______________________________________________
>>>>> SIP Express Router (SER) and Kamailio (OpenSER) - sr-users mailing
>>>>> list
>>>>> sr-users(a)lists.sip-router.org
>>>>>
http://lists.sip-router.org/cgi-bin/mailman/listinfo/sr-users
>>>>>
>>>
>>> --
>>> Yufei Tao
>>> Red Embedded
>>>
>>> This E-mail and any attachments hereto are strictly confidential and
>>> intended solely for the addressee. If you are not the intended
>>> addressee please notify the sender by return and delete the message.
>>>
>>> You must not disclose, forward or copy this E-mail or attachments to
>>> any third party without the prior consent of the sender.
>>>
>>> Red Embedded Design, Company Number 06688253 Registered in England:
>>> The Waterfront, Salts Mill Rd, Saltaire, BD17 7EZ
>>>
>
> --
> Yufei Tao
> Red Embedded
>
> This E-mail and any attachments hereto are strictly confidential and
> intended solely for the addressee. If you are not the intended
> addressee please notify the sender by return and delete the message.
>
> You must not disclose, forward or copy this E-mail or attachments to
> any third party without the prior consent of the sender.
>
> Red Embedded Design, Company Number 06688253 Registered in England:
> The Waterfront, Salts Mill Rd, Saltaire, BD17 7EZ
>
--
Yufei Tao
Red Embedded
This E-mail and any attachments hereto are strictly confidential and intended solely for
the addressee. If you are not the intended addressee please notify the sender by return
and delete the message.
You must not disclose, forward or copy this E-mail or attachments to any third party
without the prior consent of the sender.
Red Embedded Design, Company Number 06688253 Registered in England: The Waterfront, Salts
Mill Rd, Saltaire, BD17 7EZ