High availability design with Kamailio

List overview All Threads
Download

newer

older

Re: [SR-Users] Kamailio...

Shared memory fragmentation

Reinhard, Carel (GD)

5 Jun 2012 5 Jun '12

5:47 a.m.

Dear all, hope to get some information from you guys! For a redundant service we have to build up SIP proxy/registrar server in a high available scenario. Therefore, we will have 2 SIP proxy/registrar on two completely independent servers. They should be SIP dialog stateful and replicate all the session/call states between each other. If one server fails, the other should have all the SIP dialog information for call handling and accounting. How can this done in a stable and reliable way? Is this feature mature enough to support enterprise requirements? The only information I have is to build it up with the DISPATCHER module, but it seems more to be a stateless load-balancer instead of a HA module. Who has experience with HA failover designs with Kamailio and can give me some hints? Thank you very much in advance Kind regards Carel Carel Reinhard Security and VoIP Engineer SRG SSR Switzerland Telefon direkt +41 31 350 94 07 Carel.Reinhard@srgssr.ch<mailto:Carel.Reinhard@srgssr.ch>

Attachments:

attachment.html (text/html — 6.6 KB)

Show replies by date

Carsten Bock

11 Jun 11 Jun

6:32 a.m.

Hi Carel, this is a rather complex question with no simple yes or no answer. You can use the dialog module with no "in-memory" storage of dialog data and leave it to the database to do the replication. I think postgres has Multi-Master Replication, i am not sure about MySQL. At a major german telco operator, we did something similar: We had two servers, both with different addresses and an Alias for the other server. In that case, we had two loadbalancers (logically one in a HA setup) in front, who would do the failover in case. Worked fine. Carsten 2012/6/5 Reinhard, Carel (GD) <Carel.Reinhard(a)srgssr.ch>ch>:

...

-- Carsten Bock CEO (Geschäftsführer) ng-voice GmbH Schomburgstr. 80 D-22767 Hamburg / Germany http://www.ng-voice.com mailto:carsten@ng-voice.com Mobile +49 179 2021244 Office +49 40 34927219 Fax +49 40 34927220 Sitz der Gesellschaft: Hamburg Registergericht: Amtsgericht Hamburg, HRB 120189 Geschäftsführer: Carsten Bock Ust-ID: DE279344284 Hier finden Sie unsere handelsrechtlichen Pflichtangaben: http://www.ng-voice.com/imprint/

SamyGo

6:48 a.m.

...

Dear all, hope to get some information from you guys! For a redundant service we have to build up SIP proxy/registrar server

in a

high available scenario. Therefore, we will have 2 SIP proxy/registrar on two completely independent servers. They should be SIP dialog stateful

and

replicate all the session/call states between each other. If one server fails, the other should have all the SIP dialog information for call handling and accounting. How can this done in a stable and reliable way?

this feature mature enough to support enterprise requirements? The only information I have is to build it up with the DISPATCHER module, but it seems more to be a stateless load-balancer instead of a HA module. Who has experience with HA failover designs with Kamailio and can give me some hints? Thank you very much in advance Kind regards Carel Carel Reinhard Security and VoIP Engineer SRG SSR Switzerland Telefon direkt +41 31 350 94 07 Carel.Reinhard(a)srgssr.ch _______________________________________________ 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

Daniel-Constantin Mierla

7:38 a.m.

...

Hi, I personally think that there needs to be an official wiki page giving details about a basic redundant/HA server setup. A lot of people need this on regular basis. So I request forum members and contributors to share their guidelines on this. Thanks, Sammy Go. On Mon, Jun 11, 2012 at 3:32 PM, Carsten Bock <carsten(a)ng-voice.com <mailto:carsten@ng-voice.com>> wrote: Hi Carel, this is a rather complex question with no simple yes or no answer. You can use the dialog module with no "in-memory" storage of dialog data and leave it to the database to do the replication. I think postgres has Multi-Master Replication, i am not sure about MySQL. At a major german telco operator, we did something similar: We had two servers, both with different addresses and an Alias for the other server. In that case, we had two loadbalancers (logically one in a HA setup) in front, who would do the failover in case. Worked fine. Carsten 2012/6/5 Reinhard, Carel (GD) <Carel.Reinhard(a)srgssr.ch <mailto:Carel.Reinhard@srgssr.ch>>:

Dear all, hope to get some information from you guys! For a redundant service we have to build up SIP proxy/registrar

server in a

high available scenario. Therefore, we will have 2 SIP

proxy/registrar on

two completely independent servers. They should be SIP dialog

stateful and

replicate all the session/call states between each other. If one

server

fails, the other should have all the SIP dialog information for call handling and accounting. How can this done in a stable and

reliable way? Is

this feature mature enough to support enterprise requirements? The only information I have is to build it up with the

DISPATCHER module,

but it seems more to be a stateless load-balancer instead of a

HA module.

Who has experience with HA failover designs with Kamailio and

can give me

some hints? Thank you very much in advance Kind regards Carel Carel Reinhard Security and VoIP Engineer SRG SSR Switzerland Telefon direkt +41 31 350 94 07 <tel:%2B41%2031%20350%2094%2007> Carel.Reinhard(a)srgssr.ch <mailto:Carel.Reinhard@srgssr.ch> _______________________________________________ SIP Express Router (SER) and Kamailio (OpenSER) - sr-users

mailing list

sr-users(a)lists.sip-router.org <mailto:sr-users@lists.sip-router.org> http://lists.sip-router.org/cgi-bin/mailman/listinfo/sr-users

-- Carsten Bock CEO (Geschäftsführer) ng-voice GmbH Schomburgstr. 80 D-22767 Hamburg / Germany http://www.ng-voice.com mailto:carsten@ng-voice.com <mailto:carsten@ng-voice.com> Mobile +49 179 2021244 <tel:%2B49%20179%202021244> Office +49 40 34927219 <tel:%2B49%2040%2034927219> Fax +49 40 34927220 <tel:%2B49%2040%2034927220> Sitz der Gesellschaft: Hamburg Registergericht: Amtsgericht Hamburg, HRB 120189 Geschäftsführer: Carsten Bock Ust-ID: DE279344284 Hier finden Sie unsere handelsrechtlichen Pflichtangaben: http://www.ng-voice.com/imprint/ _______________________________________________ SIP Express Router (SER) and Kamailio (OpenSER) - sr-users mailing list sr-users(a)lists.sip-router.org <mailto:sr-users@lists.sip-router.org> http://lists.sip-router.org/cgi-bin/mailman/listinfo/sr-users _______________________________________________ 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

-- Daniel-Constantin Mierla - http://www.asipto.com http://twitter.com/#!/miconda - http://www.linkedin.com/in/miconda Kamailio Advanced Training, Seattle, USA, Sep 23-26, 2012 - http://asipto.com/u/katu Kamailio Practical Workshop, Netherlands, Sep 10-12, 2012 - http://asipto.com/u/kpw

SamyGo

7:46 a.m.

Sir , I always do a thorough searching through the user's mailing list here., and I know I still find it difficult thinking about creating a cluster of Kamailio servers. If only there could be basic instructions and specially role of new version modules(may be *dmq*) to create a telco-grade implementation would be a great resource. Thats just my suggestion, I am trying to setup a clustered service and still no luck. Only heartbeat works which is pretty basic, anything on top of heartbeat to make it application level redundant (active/passive OR active/active) like OCF resources and their documentation is never published or documented by anyone so far ! Thanks Sammy Go. On Mon, Jun 11, 2012 at 4:38 PM, Daniel-Constantin Mierla <miconda(a)gmail.com

...

wrote:

...

Hello, the wiki is open for everyone to contribute: * http://www.kamailio.org/wiki/ You need to make yourself an account, the anonymous posting based on CAPTCHA was disabled due to spam bots easily breaking CAPTCHA system and posting spam. If someone takes the time to search a bit through the archive, probably can find useful information to add to such page, then perhaps other people will come and contribute more. The mailing list is full of useful information, the problem is that not many people are digesting it and write wiki pages ... most of the content is by core developers, but they have other tasks as well, so a hand from user community will be more useful. Cheers, Daniel On 6/11/12 12:48 PM, SamyGo wrote: Hi, I personally think that there needs to be an official wiki page giving details about a basic redundant/HA server setup. A lot of people need this on regular basis. So I request forum members and contributors to share their guidelines on this. Thanks, Sammy Go. On Mon, Jun 11, 2012 at 3:32 PM, Carsten Bock <carsten(a)ng-voice.com>wrote;wrote:

Dear all, hope to get some information from you guys! For a redundant service we have to build up SIP proxy/registrar server

in a

high available scenario. Therefore, we will have 2 SIP proxy/registrar

two completely independent servers. They should be SIP dialog stateful

and

way? Is

this feature mature enough to support enterprise requirements? The only information I have is to build it up with the DISPATCHER

module,

but it seems more to be a stateless load-balancer instead of a HA

module.

Who has experience with HA failover designs with Kamailio and can give

some hints? Thank you very much in advance Kind regards Carel Carel Reinhard Security and VoIP Engineer SRG SSR Switzerland Telefon direkt +41 31 350 94 07 Carel.Reinhard(a)srgssr.ch _______________________________________________ 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

-- Carsten Bock CEO (Geschäftsführer) ng-voice GmbH Schomburgstr. 80 D-22767 Hamburg / Germany http://www.ng-voice.com mailto:carsten@ng-voice.com Mobile +49 179 2021244 <%2B49%20179%202021244> Office +49 40 34927219 <%2B49%2040%2034927219> Fax +49 40 34927220 <%2B49%2040%2034927220> Sitz der Gesellschaft: Hamburg Registergericht: Amtsgericht Hamburg, HRB 120189 Geschäftsführer: Carsten Bock Ust-ID: DE279344284 Hier finden Sie unsere handelsrechtlichen Pflichtangaben: http://www.ng-voice.com/imprint/ _______________________________________________ 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

_______________________________________________ SIP Express Router (SER) and Kamailio (OpenSER) - sr-users mailing listsr-users@lists.sip-router.orghttp://lists.sip-router.org/cgi-bin/mailman/listinfo/sr-users -- Daniel-Constantin Mierla - http://www.asipto.comhttp://twitter.com/#!/miconda - http://www.linkedin.com/in/miconda Kamailio Advanced Training, Seattle, USA, Sep 23-26, 2012 - http://asipto.com/u/katu Kamailio Practical Workshop, Netherlands, Sep 10-12, 2012 - http://asipto.com/u/kpw

Juha Heinanen

7:56 a.m.

SamyGo writes:

...

Thats just my suggestion, I am trying to setup a clustered service and still no luck. Only heartbeat works which is pretty basic, anything on top of heartbeat to make it application level redundant (active/passive OR active/active) like OCF resources and their documentation is never published or documented by anyone so far !

SamyGo

8:01 a.m.

Thanks Juha :) Really appreciate that. Would you please like to contribute the whole procedure so that everyone knows whats happening here !! As soon as I finish with this one *successfully* I will write about it in my blog, so anything that leads to its successful implementation would benefit alot of other folks. Thanks again BR Sammy On Mon, Jun 11, 2012 at 4:56 PM, Juha Heinanen <jh(a)tutpro.com> wrote:

...

SamyGo writes:

Thats just my suggestion, I am trying to setup a clustered service and still no luck. Only heartbeat works which is pretty basic, anything on

top

of heartbeat to make it application level redundant (active/passive OR active/active) like OCF resources and their documentation is never published or documented by anyone so far !

there is nothing special in kamalio crm primitive definition, for example: primitive sip-proxy lsb:kamailio op monitor interval=15s timeout=10s \ start-delay=60s net has many examples on how you combine such a primitive to a group with a virtual-ip primitive. -- juha _______________________________________________ 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

Juha Heinanen

8:15 a.m.

SamyGo writes:

...

Really appreciate that. Would you please like to contribute the whole procedure so that everyone knows whats happening here !!

i'm not able write a tutorial on how to setup and use heartbeat/pacemaker. virtual-ip and sip-proxy could be configured something like this (i haven't tested it): primitive virtual-ip ocf:heartbeat:IPaddr2 params ip=192.75.101.10 broadcast=192.75.101.255 cidr_netmask=24 op monitor interval=15s primitive sip-proxy lsb:sip-proxy op monitor interval=15s timeout=15s start-delay=60s order virtual-ip-before-sip-proxy mandatory: virtual-ip:start sip-proxy colocation sip-proxy-on-virtual-ip inf: sip-proxy virtual-ip -- juha

SamyGo

8:20 a.m.

OK thanks, I'm trying this all for myself. once I try this, test this, and get something done I will post the tutorial for all.* I may need help in this.* Regards, Sammy On Mon, Jun 11, 2012 at 5:15 PM, Juha Heinanen <jh(a)tutpro.com> wrote:

...

SamyGo writes:

Really appreciate that. Would you please like to contribute the whole procedure so that everyone knows whats happening here !!

SamyGo

12 Jun 12 Jun

9:05 a.m.

Hi again, Some progress.. I have configured heartbeat and with the crm option to set kamailio as resource i.e crm(test-conf)configure# *primitive Kamailio lsb::kamailio op monitor interval=15s* I can see that if I put the primary node on standby the kamailio instance starts on secondary node ! but this isn't working other way around e.g If Kamailio crashes the secondary node doesn't do anything. I've read people that they used sipsak and some piece of code that pings Kamailio and if it fails to respond it triggers to heartbeat. Is there any help on this ! Regards, Sammy G. On Mon, Jun 11, 2012 at 5:20 PM, SamyGo <govoiper(a)gmail.com> wrote:

...

OK thanks, I'm trying this all for myself. once I try this, test this, and get something done I will post the tutorial for all.* I may need help in this. * Regards, Sammy On Mon, Jun 11, 2012 at 5:15 PM, Juha Heinanen <jh(a)tutpro.com> wrote:

SamyGo writes:

Really appreciate that. Would you please like to contribute the whole procedure so that everyone knows whats happening here !!

SamyGo

9:11 a.m.

Infact if kamailio goes down heartbeat restarts it again !! how do I stop heartbeat doing it and let the other node take over the resource! On Tue, Jun 12, 2012 at 6:05 PM, SamyGo <govoiper(a)gmail.com> wrote:

...

SamyGo writes:

Really appreciate that. Would you please like to contribute the whole procedure so that everyone knows whats happening here !!

4595

days inactive

4602

days old

sr-users@lists.kamailio.org

Manage subscription

10 comments

5 participants

tags (0)

participants (5)

Carsten Bock
Daniel-Constantin Mierla
Juha Heinanen
Reinhard, Carel (GD)
SamyGo