Hello,
I currently have Freeswitch acting as a B2BUA, handling registrations,
routing, etc for a prototype voice application I have built.
Now I'm at the point where it's time to actually scale it out, and I'm
looking at a few different options for the SIP proxy + registration later.
Essentially,
Hey all,
I'm looking to put Kamailio behind a TCP load balancer that is SIP-unaware.
My application is deployed in AWS and I'm tying to place Kamailio behind an
ELB.
For the most part, everything is fine. For my specific implementation I'm
disabling UDP as a signaling transport and using only TLS
is also quite fast.
> Cheers,
> Daniel
>
>
> On 23/06/16 04:32, Colin Morelli wrote:
>
> Hey all,
>
> I'm looking to put Kamailio behind a TCP load balancer that is
> SIP-unaware. My application is deployed in AWS and I'm tying to place
> Kamailio behind an
Hey all,
I'm using a combination of http_async_client and rtjson to query my API and
retrieve a JSON target route set for an incoming SIP request that Kamailio
will forward to. The HTTP portion of it works great. I'm able to hit my API
and get back a JSON document.
Additionally, the rtjson parsin
Sorry all - I've been staring a kamailio config files for too long.
The issue was obvious when I read log messages more closely: the
route[REROUTE] needed to be failure_route[REROUTE]. Everything works fine
now .
Best,
Colin
On Fri, Jun 24, 2016 at 3:01 PM Colin Morelli
wrote:
>
Hey all,
When using Kamailio at the edge - what's the best practice around how to
advertise your Record-Route? I assume it's going to involve the use of a
double-RR with both the public and private IPs. However, I'm running in AWS
where the host doesn't have two interfaces with both a public and p
Hey Alex,
Thanks for the response. This is the AWS scenario where there's a 1:1 NAT
from the public to private IP.
I've got as far as figuring out how to advertise the public IP. But, when I
forward the request to another node inside the cluster, I assume I want to
double-RR that request so that
Thanks for the quick reply!
Binding to two SIP ports isn't out of the question (though I'd like to
avoid it if possible).
However, with this approach, I assume somewhere I must have to instruct
Kamailio which outbound interface to use (i.e. tell it to use 5080 for
forwarded requests to internal h
Awesome, thank you.
If I were to try to avoid opening another point, would it be sensible to
call record_route_advertised_address() with the advertised address twice
manually (once for the inbound and outbound legs with the appropriate IPs
for each)? Internally I assume Kamailio's loose_route() wo
Alright, I'll give both approaches a shot and see what comes up.
Thanks for the fast response time, Alex!
Best,
Colin
On Sat, Jun 25, 2016 at 7:07 PM Alex Balashov
wrote:
> It can work, but it's more trouble than the other approach, which is
> essentially automagic.
>
> -- Alex
>
> --
> Princi
Hey all,
Back with more questions.
I'm using Kamailio to make an HTTP call to my API to perform authentication
and message routing. Currently, I'm trying to build up the post body that I
send to my API to make those decisions.
I've cherry picked a few of the headers that are important in my rout
nd pushing those up.
Thanks all!
Best,
Colin
On Mon, Jun 27, 2016 at 7:49 AM Olle E. Johansson wrote:
>
> > On 26 Jun 2016, at 22:29, Colin Morelli wrote:
> >
> > Hey all,
> >
> > Back with more questions.
> >
> > I'm using Kamailio to make an
I suppose there's a lot of subjectivity here - and it greatly depends on
your configuration - but at least for my use case I don't quite agree with
that statement. My API is already the component performing authentication
and making routing decisions anyway, which means Kamailio is going to send
th
Maybe I'm missing something about the core infrastructure of Kamailio that
makes this impossible, but why does it seem like nobody wants to run
multiple Kamailio load balancers in a cluster? sip.yourcompany.com can have
A/SRV records pointing to multiple IP addresses of separate Kamailio
instances.
Daryn,
That response was more general, not necessarily directed at you!
DNS-based load balancing has always been problematic for clients. They tend
to not properly balance across SRV records, or failover to secondary A
records.
However, I think the best solution would be something like what Dani
Hey all,
I'm running a cluster of Kamailio instances as a proxy/registrar for
another cluster of Freeswitch instances. I'm using http_async_client to
make HTTP queries to my API to fetch credentials on auth challenges.
Kamailio performs generating the header, and validating the result based on
the
Hey list,
So I'm using the outbound module to ensure subsequent requests are
delivered over an active TCP connection established by the mobile client.
However, now I'm trying to add support for automatically responding to
network changes (WiFi <-> LTE), and it's creating problems.
Primarily, the
orking.
Best,
Colin
On Fri, Jul 8, 2016 at 10:32 AM Colin Morelli
wrote:
> Hey list,
>
> So I'm using the outbound module to ensure subsequent requests are
> delivered over an active TCP connection established by the mobile client.
> However, now I'm trying to add supp
Hey all,
I'm using Kamailio as an outbound edge proxy for websocket connections.
When my registrar calls out to Kamailio to forward to a websocket
connection that has since been killed, rather than just entering the
failure branch, it throws a few errors:
5(24) WARNING: [msg_translator.c:2760]:
eter gets what you are looking for:
>
> -
> https://www.kamailio.org/docs/modules/stable/modules/tm.html#tm.p.failure_exec_mode
>
> Cheers,
> Daniel
>
> On 14/07/16 19:34, Colin Morelli wrote:
>
> Hey all,
>
> I'm using Kamailio as an outbound edge proxy f
Zaka,
I could be wrong here but I don't think you ever actually have a "listen"
line for MY_WS_ADDR.
I believe you have a typo, as you have listen=MY_IP_ADDR twice, once within
the guard for WITH_WEBSOCKETS. Replace the one inside the if with
MY_WS_ADDR and I think your problem should be resolved
ced the lack of *MY-WS_ADDRESS*
>
> Last, but not the least, in a browser supporting WEBRTC (FF, CHROME) do we
> still need SIP ML5as on: https://www.doubango.org/sipml5/
>
> If not, I wonder how will the WS or WSS URL post the login credentials to
> the KAMAILIO WEBRTC s
If you're using Kamailio as a registrar, then it would make the most sense
to also use it as your outbound proxy for Asterisk.
This would mean whenever Asterisk needs to dial an extension, it would
instead make a SIP call to your Kamailio instance which would then perform
the lookup, forking, and
Hey all,
I've got two layers of Kamailio proxies running. One set of edge proxies
that are parking outbound connections and doing load balancing to a set of
registrar/proxies. This is working well with one exception:
I can't seem to get t_load_contacts/t_next_contacts working correctly for
handli
ke a
case that t_load_contacts *should* handle itself, no?
Best,
Colin
On Sun, Jul 24, 2016 at 12:35 PM Juha Heinanen wrote:
> Colin Morelli writes:
>
> > 8(27) DEBUG: tm [t_serial.c:191]: t_load_contacts(): nr_branches is 0
> > 8(27) DEBUG: tm [t_serial.c:194]: t_load_contacts()
unclear, but this was a request for
clarification, not a bug report - though I will try to make a PR for the
outbound documentation page when I get this working.
Best,
Colin
On Sun, Jul 24, 2016 at 12:51 PM Juha Heinanen wrote:
> Colin Morelli writes:
>
> > When there's one reg
Luke,
This is actually the behavior I would expect, though admittedly I've never
tried to rely on received= parameter routing for requests (I wasn't aware
that was something that should be supported).
However, you're probably better of addressing your NAT issues in a
different part of your reques
Hey Jonathan,
You can also use the Jansson module to create the JSON request. For
example, in one of my configs, I have a route that looks basically like:
$var(params) = $null;
jansson_set("string", "request.uri", "$ru", "$var(params)");
jansson_set("string", "request.method", "$rm", "$var(params
Hey all,
Looking to figure out the best way to allow TCP connections to stay alive
for NAT'd clients, however, to protect against people just opening TCP
connections to the server, I'm hoping to only keepalive TCP connections for
connections that have sent an INVITE and received a 200 OK.
Does th
> You can do that when you get the INVITE, before sending a negative
> response from kamailio.cfg. Or, if you relay the invite, set a failure
> route for it and do the operation there.
>
> Cheers,
> Daniel
>
> On 07/11/16 23:26, Colin Morelli wrote:
>
> Hey all
Hey all,
I know Kamailio's registrar module has a max_contacts parameter that will
limit the number of active contacts for an AOR. However, is there any
mechanism to control what happens when that number is hit?
When using a web client where every page refresh registers a new contact
with Kamaili
ot only for web clients?
>
> On December 1, 2016 5:54:06 PM EST, Colin Morelli
> wrote:
> >Hey all,
> >
> >I know Kamailio's registrar module has a max_contacts parameter that
> >will
> >limit the number of active contacts for an AOR. However, is there any
>
Hey all,
Sorry for two questions out at the same time!
I'm using kamailio in front of FS, and I'm about to enable IPv6, but I have
a question that I'm not entirely clear on.
What's the proper way to forward traffic using the same interface that it
came in on? Specifically, if a client connects t
stration from a device will
> replace its own old contact, even if it has a different contact address.
>
> Cheers,
> Daniel
>
> On 01/12/2016 23:59, Colin Morelli wrote:
>
> Hey Alex - not sure I'm quite following what you mean. The bindings are
> only for web clients.
Can't answer whether or not it's possible as I haven't tried it - but,
Consider how this will impact the negotiation between clients and servers.
If your B2BUA thinks calls are coming in from IPV4 (instead of IPV6), then
it may not offer IPV6 in the invite. Similarly, if the client thinks it's
con
Hey all,
Trying to debug an issue with canceling an invite. I have two different
types of clients. On one client, canceling an invite works correctly. With
the other client, it t_check_trans fails.
Both clients show the same request/responses:
On Client A:
-> INVITE
<- 183
-> PRACK
<- 200
-> UPD
, which caused the transaction to not match. Through some
configuration changes in PJSIP I was able to work around this.
Best,
Colin
On Sat, Apr 8, 2017 at 7:56 PM, Colin Morelli
wrote:
> Hey all,
>
> Trying to debug an issue with canceling an invite. I have two different
> types of
37 matches
Mail list logo