19 Feb
2005
19 Feb
'05
11:13 p.m.
i just committed some patches to lcr module that introduced two new
functions: load_contacts() and next_contacts(). they can be used to
implement sequential forking (see README for details).
if all contacts in the destination set have the same qvalue,
load_contacts() and next_contacts() don't do anything, just check that
nothing needs to be done.
-- juha
20 Feb
20 Feb
1:42 a.m.
Juha,
Can sequential forking (mutiple contacts) be used with multiple
gateways? In other words if next_gw() is used from a failure_route
block to try failover gateways, can I also use the next_contacts() for
serial forking?
If so, can you tell me how I would structure my failure_route or other
necessary code blocks?
Regards,
Paul
On Sat, 19 Feb 2005 23:13:56 +0200, Juha Heinanen <jh(a)lohi.tutpro.com> wrote:
i just committed some patches to lcr module that
introduced two new
functions: load_contacts() and next_contacts(). they can be used to
implement sequential forking (see README for details).
if all contacts in the destination set have the same qvalue,
load_contacts() and next_contacts() don't do anything, just check that
nothing needs to be done.
-- juha
_______________________________________________
Serusers mailing list
serusers(a)lists.iptel.org
http://lists.iptel.org/mailman/listinfo/serusers
1:58 a.m.
Java Rockx writes:
Can sequential forking (mutiple contacts) be used with
multiple
gateways? In other words if next_gw() is used from a failure_route
block to try failover gateways, can I also use the next_contacts() for
serial forking?
as the name says, next_gw() gives you only one next gateway. there is
thus no need for serial forking.
If so, can you tell me how I would structure my
failure_route or other
necessary code blocks?
basically you call next_gw() and if that succeeds you set failure route
again and call t_relay.
-- juha
2:07 a.m.
Yeah, I saw that in the README file. But I guess I was asking if
next_contacts() could be used to implement a type of "follow-me"
service in addition to the PSTN failover functionality that next_gw()
provides.
Regards,
Paul
On Sun, 20 Feb 2005 01:58:00 +0200, Juha Heinanen <jh(a)lohi.tutpro.com> wrote:
Java Rockx writes:
Can sequential forking (mutiple contacts) be used
with multiple
gateways? In other words if next_gw() is used from a failure_route
block to try failover gateways, can I also use the next_contacts() for
serial forking?
as the name says, next_gw() gives you only one next gateway. there is
thus no need for serial forking.
If so, can you tell me how I would structure my
failure_route or other
necessary code blocks?
basically you call next_gw() and if that succeeds you set failure route
again and call t_relay.
-- juha
2:22 a.m.
Java Rockx writes:
Yeah, I saw that in the README file. But I guess I was
asking if
next_contacts() could be used to implement a type of "follow-me"
service in addition to the PSTN failover functionality that next_gw()
provides.
load_contacts() loads whatever contacts are in the destination set
usually after enum or location lookup or perhaps after some avpops
calls. next_contacts() calls then "follow" those contacts and you may
get what you mean by a "follow me" service.
-- juha
1 Mar
1 Mar
5 p.m.
New subject: LCR Module (Was: Re: [Serusers] sequential forking support)
On Sun, 20 Feb 2005, Juha Heinanen wrote:
Java Rockx writes:
Hi,
Been trying out the LCR module from CVS (head, checkout today) and I can't
get it to work when the first gateway fails.
Setup:
mysql> select * from lcr;
+--------+----------+--------+----------+
| prefix | from_uri | grp_id | priority |
+--------+----------+--------+----------+
| | % | 1 | 0 |
+--------+----------+--------+----------+
mysql> select * from gw;
+------------------+-----------+------+--------+
| gw_name | ip_addr | port | grp_id |
+------------------+-----------+------+--------+
| sip-gw.swip.net | 247264386 | 5060 | 1 |
| sip2-gw.swip.net | 247264385 | 5060 | 1 |
+------------------+-----------+------+--------+
mysql> select * from gw_grp;
+--------+---------------+
| grp_id | grp_name |
+--------+---------------+
| 1 | PSTN Gateways |
+--------+---------------+
sip2-gw.swip.net doesn't exist i.e. requests to it will timeout.
In route{} I have:
if (!load_gws()) {
sl_send_reply("500", "Server Internal Error - Cannot load
gateways");
break;
};
And then calls to PSTN end up in route[4]
route[4]{
if (method=="INVITE" | method=="BYE") {
xlog("L_INFO","Routeblock 4 - Call not to customer, to
gateways\n");
};
t_on_failure("1");
if (!next_gw()) {
sl_send_reply("503", "Service not available - No
gateways");
break;
};
if (!t_relay()) {
sl_reply_error();
};
}
failure_route[1] {
xlog("L_INFO","Failureroute 1 - Get next gateway from
LCR\n");
if (!next_gw()) {
xlog("L_INFO","Failureroute 1 - No more gateways
available\n");
t_reply("503", "Service not available - No more
gateways");
break;
} else {
t_on_failure("1");
t_relay();
};
}
When I call and LCR selects the working gateway first it works ok, but
when it selects the gateway that doesn't exist I get this.
0(7028) Tue Mar 1 08:39:33 2005 - <null> - INVITE
Call-ID: 00036bc3-7aa52432-43a4bc4f-121393fe(a)130.244.194.233
From: sip:0856204081@sip-corporate1.testdomain.com
To: sip:0890510@sip-corporate1.testdomain.com
0(7028) Routeblock 2 - Calls from customers
0(7028) Call from customer: Test
0(7028) Call passed A-number check
0(7028) Routeblock 3 - To customer?
0(7028) Routeblock 4 - Call not to customer, to gateways
1(7032) Failureroute 1 - Get next gateway from LCR
1(7032) BUG: qm_free: bad pointer 0xb57a1bf0 (out of memory block!) - aborting
0(7028) child process 7032 exited by a signal 6
0(7028) core was generated
0(7028) INFO: dont_fork turned on, living on
2(7034) INFO: signal 15 received
0(7028) Thank you for flying ser
I have restarted ser after changing tables in mysql.
Any hints?
Best regard,
Thomas Björklund
If so, can you tell me how I would structure my
failure_route or other
necessary code blocks?
basically you call next_gw() and if that succeeds you set failure route
again and call t_relay.
2 Mar
2 Mar
8:28 a.m.
New subject: LCR Module (Was: Re: [Serusers] sequential forking support)
thomas,
i looked at next_gw and it allocated memory for uri before called
do_action to append the branch and then it free the memory. so i didn't
see anything obvious tha could cause your core dump.
can you start ser something like
ser -u root -g root -w /tmp
so that it produces a core dump. then run
gdb /usr/sbin/ser /tmp/core
and give it "where" command and send me the result.
-- juha
10:59 a.m.
New subject: LCR Module (Was: Re: [Serusers] sequential forking support)
On Wed, 2 Mar 2005, Juha Heinanen wrote:
thomas,
i looked at next_gw and it allocated memory for uri before called
do_action to append the branch and then it free the memory. so i didn't
see anything obvious tha could cause your core dump.
And my configuration in failure_route is ok?
failure_route[1] {
if (!next_gw()) {
t_reply("503", "Service not available - No more
break;
} else {
t_on_failure("1");
t_relay();
};
}
I don't need append_branch or something?
can you start ser something like
ser -u root -g root -w /tmp
so that it produces a core dump. then run
gdb /usr/sbin/ser /tmp/core
and give it "where" command and send me the result.
-- juha
(gdb) where
#0 0x008bfcef in raise () from /lib/tls/libc.so.6
#1 0x008c14f5 in abort () from /lib/tls/libc.so.6
#2 0x08081406 in qm_free (qm=0x80fad40, p=0xb579dbec, file=0x9cc898
"¤G\023", func=0x809e9b7 "free_lump", line=375) at mem/q_malloc.c:445
#3 0x080532cc in free_lump (lmp=0xb579dbcc) at data_lump.c:377
#4 0x003b7223 in free_via_clen_lump (list=0x3d8eb4) at fix_lumps.h:73
#5 0x003b5d0c in print_uac_request (t=0x3d0664, i_req=0x3d8c20, branch=0,
uri=0x0, len=0xbfff907c, send_sock=0x0, proto=PROTO_NONE) at t_fwd.c:109
#6 0x003b613e in add_uac (t=0xb579c778, request=0x3d8c20, uri=0xbfff9118,
next_hop=0xbfff9118, proxy=0x813983c, proto=1) at t_fwd.c:228
#7 0x003b696f in t_forward_nonack (t=0xb579c778, p_msg=0x3d8c20,
proxy=0x0, proto=0) at t_fwd.c:453
#8 0x003be00e in w_t_relay (p_msg=0x3d8c20, _foo=0x0, _bar=0x0) at
tm.c:886
#9 0x0804f1aa in do_action (a=0x81399ec, msg=0x3d8c20) at action.c:611
#10 0x08050a6c in run_actions (a=0x81399ec, msg=0x3d8c20) at action.c:719
#11 0x0804f0c9 in do_action (a=0x8139a3c, msg=0x3d8c20) at action.c:601
#12 0x08050a6c in run_actions (a=0x8139a3c, msg=0x3d8c20) at action.c:719
#13 0x003c736e in run_failure_handlers (t=0xb579c778, rpl=0x0, code=0) at
t_reply.c:629
#14 0x003c6713 in t_should_relay_response (Trans=0xb579c778, new_code=408,
branch=0, should_store=0xbfff9638, should_relay=0xbfff963c,
cancel_bitmap=0x0, reply=0x0)
at t_reply.c:772
#15 0x003c2daf in relay_reply (t=0xb579c778, p_msg=0xffffffff, branch=0,
msg_status=408, cancel_bitmap=0xbfff96a8) at t_reply.c:994
#16 0x003b9413 in final_response_handler (fr_tl=0xffffff9c) at timer.c:251
#17 0x003b8d3d in timer_routine (ticks=7, attr=0x0) at timer.c:952
#18 0x0807adbb in timer_ticker () at timer.c:146
#19 0x0805bda1 in main_loop () at main.c:873
#20 0x0805d21c in main (argc=-1251896308, argv=0x0) at main.c:1568
Best regards,
Thomas Björklund
11:16 a.m.
New subject: LCR Module (Was: Re: [Serusers] sequential forking support)
thomas,
in your gdb "where", i don't see next_gw function call. so perhaps your
crash has nothing to do with lcr module? do other people have any
ideas?
And my configuration in failure_route is ok?
failure_route[1] {
if (!next_gw()) {
t_reply("503", "Service not available - No more
break;
} else {
t_on_failure("1");
t_relay();
};
}
your configuration looks ok, but you may want to limit the replies for
which you try another wg. for example, if you get "busy", it won't help
to try another gw.
also, before you call t_relay from route block, set fr_timer to a low
value.
-- juha
7205
days inactive
7216
days old
8 comments
3 participants
participants (3)
-
Java Rockx -
Juha Heinanen -
Thomas Björklund