[OpenSIPS-Users] load_balance not using group id in failure route

Tue Jan 3 15:45:55 CET 2012

Hi Răzvan,

Okay, thank you!
I'll investigate what I can do with the dispatcher.

Steven

From: users-bounces at lists.opensips.org [mailto:users-bounces at lists.opensips.org] On Behalf Of Razvan Crainea
Sent: dinsdag 3 januari 2012 15:41
To: users at lists.opensips.org
Subject: Re: [OpenSIPS-Users] load_balance not using group id in failure route

Hi, Steven!

This is not strange, the usual scenario is to do failover within the gateways from the same group. My question is that if you have only one resource, why don't you use the dispatcher[1] module? I think it will meet better your requirements, as load balancer is usually used when you want to dispatch requests based on multiple resources. Also, there you can disable the automatic failover behavior.

[1] http://www.opensips.org/html/docs/modules/1.7.x/dispatcher.html

Regards,

--

Răzvan Crainea

OpenSIPS Developer

On 01/03/2012 04:31 PM, Steven Lam, KeenSystems B.V. wrote:
Hi Răzvan,

Thank you for your answer!
The table was just for illustrating the problem as simple as possible...

What I want to achieve is this: 3 servers for everyday use and 2 servers for fallback, so simple RURI rewriting won't do the trick :-(

Can you tell me what the problem is with resetting the AVPs?

It is good to hear it is not a bug, but this makes me thinking... in this example the resources in all groups are "all". If I change the one from group 2 to "foobar" and then call load_balance in the failure_route like "load_balance("2","foobar")", I get a line telling me resource "foobar" is not found in group 1???

So in a failure_route load_balance does care about the resource and does NOT care about the group number, is this not strange?

Steven

From: users-bounces at lists.opensips.org<mailto:users-bounces at lists.opensips.org> [mailto:users-bounces at lists.opensips.org] On Behalf Of Razvan Crainea
Sent: dinsdag 3 januari 2012 14:36
To: users at lists.opensips.org<mailto:users at lists.opensips.org>
Subject: Re: [OpenSIPS-Users] load_balance not using group id in failure route

Hi, Steven!

This is not wrong, this is the desired behaviour in order to use the failover features.
I see that you have only one destination in group 2. Will this be the final scenario? If yes, then you shouldn't use load balance, but a simple RURI rewriting.
However, if you want to use multiple gateways from failure route, there is a hack you can do, but I don't recommend it. You can reset the lb_mask, lb_id and lb_grp AVPs just before calling the load_balance function.

Regards,

--

Răzvan Crainea

OpenSIPS Developer

On 01/03/2012 02:34 PM, Steven Lam, KeenSystems B.V. wrote:
Is nobody using this? :-(
I really want to know if I'm doing something wrong here...

Steven

From: users-bounces at lists.opensips.org<mailto:users-bounces at lists.opensips.org> [mailto:users-bounces at lists.opensips.org] On Behalf Of Steven Lam, KeenSystems B.V.
Sent: donderdag 29 december 2011 14:48
To: users at lists.opensips.org<mailto:users at lists.opensips.org>
Subject: [OpenSIPS-Users] load_balance not using group id in failure route

Hi all,

Using the load balancer on OpenSIPS 1.7.1 I found that when calling load_balance in a failure_route with a group other than the one on the initial call, the load_balance still uses the "old" group.

For example:

The table looks like this:
+----+----------+------------------+-----------+
| id | group_id | dst_uri          | resources |
+----+----------+------------------+-----------+
|  1 |        1 | sip:192.168.9.22 | all=10    |
|  2 |        1 | sip:192.168.9.27 | all=20    |
|  3 |        2 | sip:192.168.9.18 | all=10    |
+----+----------+------------------+-----------+

Script looks like this:
...
if ( load_balance("1","all") ) {
     xlog("==================> Destination is $du\n");
     t_on_failure("1");
     t_relay();
     exit;
}
...

failure_route[1] {
 lb_disable();
  if ( load_balance("2","all") ) {
    t_on_failure("1");
    xlog("==================> New destination is $du\n");
    t_relay();
  } else {
    t_reply("500","Error");
  }
}

This will result in routing a INVITE to sip:192.168.9.27, when this fails the INVITE will be routed to sip:192.168.9.22 _NOT_ 192.168.9.18. To me this looks wrong.

Is this by design? If so how can I force load_balance to use a uri from group 2 after using one from group 1?

Hope someone can give me some advice on this.

Steven

_______________________________________________

Users mailing list

Users at lists.opensips.org<mailto:Users at lists.opensips.org>

http://lists.opensips.org/cgi-bin/mailman/listinfo/users

_______________________________________________

Users mailing list

Users at lists.opensips.org<mailto:Users at lists.opensips.org>

http://lists.opensips.org/cgi-bin/mailman/listinfo/users
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.opensips.org/pipermail/users/attachments/20120103/49c70e01/attachment.htm>