Cool!


On 4 August 2010 21:15, jay binks <jaybinks@gmail.com> wrote:
Guys,
ššfurther to this interesting discussion, please check out this wiki pageš

http://wiki.freeswitch.org/wiki/Freeswitch_HA

there was an announcement at cluecon that improves Freeswitches ability to do HA-clustering.
please can you update this wiki page with the information you have learned doing the pacemaker side of this.

also keep the conversation going now, with this new information :)

Jay


2010/7/12 Steven Ayre <steveayre@gmail.com>

"As server FS work with share db, there is an impression that at switching Cluster-IP with active on
a passive, passive server can pick up session begun on an active server (the necessary data about
current sessions takes from the share db). Whether So it?
Testing has shown that at switching Cluster-IP on reserve FS, that does not process session which
have begun on first FS :("

Sorry, but that's not currently possible as the call state information is held in memory, not the shared database, and has execution threads started for processing the signalling/media which aren't duplicated on the other machine(s).

-Steve



2010/7/9 viewpoint <yurazilot1@list.ru>

Many thanks for fast and substantial answers!
Has established ipv4.ip_nonlocal_bind=1.
Now there is no necessity to reboot FS or sofia profile.

As a result it has turned out:
1) Two servers FS which kernel works with the shared db through unix ODBC
2) It Is realised Cluster-IP by installations Pacemaker
As server FS work with share db, there is an impression that at switching Cluster-IP with active on
a passive, passive server can pick up session begun on an active server (the necessary data about
current sessions takes from the share db). Whether So it?
Testing has shown that at switching Cluster-IP on reserve FS, that does not process session which
have begun on first FS :(



Thu, 8 Jul 2010 16:21:40 +0200 (CEST) ÐÉÓØÍÏ ÏÔ Vladimir Klejch <klejch+freeswitch@netbox.cz>:

>
> Hi
>
> On linux you can use /proc/sys/net/ipv4/ip_nonlocal_bind to bind to non local
> address on start of FS and then you don't need any restart of FS or
> reload of sofia profile if this addres is active on the node
>
> or you can have this floating address on dummy iface (?or lo ) šand then
> use right settings in /proc/sys/net/ipv4/conf/eth?/arp_filter,arp_ignore
> and then FS can use this address to bind on start and you don't need
> restart of FS?or relaod of sofia profile ...
>
>
>
> š š š š š š š š š š š š š š š Kleo
>
> On Thu, 8 Jul 2010, viewpoint wrote:
>
> > Hello.
> >
> > I currently have a project where I'm researching how to establish a
> > clustered platform wieth failover time ~= some milliseconds.
> >
> > At present we have: 2 identical servers FS, wieth established Pacemaker
> > (http://www.clusterlabs.org/wiki/Install#From_Source) - heartbeat. On
> > the one hand it is possible to tell that all works, as, at refusal of
> > the first server, the second server receives cluster-ip. A problem that
> > FreeSWITCH it is necessary to restart (or simply to start) that it began
> > to work with the new IP-address. As the result, a switching total time
> > makes 10-20 sec which basic part is necessary on launch FreeSwitch.
> >
> > any ideas?
> >
> > Thanks.
>
>
> --
> š š š klejch+freeswitch@netbox.cz
>


_______________________________________________
FreeSWITCH-users mailing list
FreeSWITCH-users@lists.freeswitch.org
http://lists.freeswitch.org/mailman/listinfo/freeswitch-users
UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users
http://www.freeswitch.org




--
Sincerely

Jay

_______________________________________________
FreeSWITCH-users mailing list
FreeSWITCH-users@lists.freeswitch.org
http://lists.freeswitch.org/mailman/listinfo/freeswitch-users
UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users
http://www.freeswitch.org