[jadmin] Highly Available Jabber server

[Zbyszek Żółkiewski]

http://ejabberd.jabber.ru/cisco-slb



On 10/19/06, Daryl Herzmann <akrherz at gmail.com> wrote:
>
> On 10/18/06, fulan Peng <fulanpeng at gmail.com> wrote:
> > I have setup two ejabberd servers. I have clustered its mnesia
> > database but I do not know how to configure jwchat on it. Now I set up
> > a Squid proxy on one of the machine and server as proxy for another
> > one. You may have a test on it:
> > https://breakevilaxis.org
> > https://ddint.org
>
> thanks for the email.  I've been reading some more and think I
> understand the concept of a ejabberd cluster, but there is still the
> issue of how to make it highly available in case of a cluster node
> failure.  I see options like:
>
> 1.  You have some sort of DNS solution and when a node fails, you
> remove its IP from round robin DNS.  This seems problematic for
> periods of downtime while DNS updates.
>
> 2.  You have a hardware solution that makes a single internet presence
> out of real servers in the ejabberd cluster.
>
> 3.  You use IPVS with two directors and two real servers. I currently
> do that very successfully with apache, but I need this chat solution
> to be deployed at a colo/hoster.
>
> It seems most hoster sites do not support IP failover (which IPVS)
> wants.  And I suspect my budget doesn't match my uptime requirements
> :)
>
> am I missing something obvious?
> thanks,
> daryl
> _______________________________________________
> JAdmin mailing list
> JAdmin at jabber.org
> http://mail.jabber.org/mailman/listinfo/jadmin
> FAQ: http://www.jabber.org/about/jadminfaq.shtml
> _______________________________________________
>



-- 
pozdrawiam,
Zbyszek Żółkiewski
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.jabber.org/pipermail/jadmin/attachments/20061019/78031643/attachment-0003.htm