httpd-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Tom Evans <>
Subject Re: [users@httpd] How to set up 2 or more machines running HTTPD instances to serve a same request ( in a load balanced fashion).
Date Thu, 05 Apr 2012 09:46:17 GMT
On Thu, Apr 5, 2012 at 1:22 AM, Pratapani
<> wrote:
> Hi,
> Greetings !!
> I am writing this after lot of googling. I could not get a direct answer in
> my searches.
> As I  don't have a more time to do R&D, I am sending this mail for a quick
> solution.
> I sucessfully did the following :
> --------------------------------------------
> I am able to have one HTTPD server running and 2 tomcats instances ( load
> balanced )  for  the single HTTPD server.
> I was able to sucessfully test the Session replication in case of 1 tomcat
> instance going down.
> The request is serviced by another tomcat instance and session is also
> replicated.
> I am unable to do the following :
> ---------------------------------------------
> In the above case if the machine hosting the HTTPD is down ,  the requests
> are not serviced although the tomcast instances are available.
> How do I setup HTTPD servers as load balaced. That is 2 or more machines
> serving the same address.
> There is virtual host to support a single machine serve multple address .
> Just would like to understand if there is a simple way to set up 2 or more
> machines running HTTPD instances to serve a same request ( in a load
> balanced fashion).
> Greatly appreciate your help.
> Thanks ,
> Durga Prasad

This is outside the scope of Apache tbh. We do this though.

We have a pair of servers acting as front end proxies.
Each server has two HA addresses - one server is the master on one HA
address, the other server master on the other.
Each server also has all the public IPs we intend to serve on the
loopback interface. This allows httpd to bind to the correct IP
addresses on each box.
Our edge routers forward traffic for those public IPs in a round robin
format via the two HA addresses.
If either server is down, the request is handled by the other server,
due to the magic of HA.

The technologies we use here:

All servers run FreeBSD. Round robin load balancing happens by PF (I
think). We use CARP addresses for HA - FreeBSD version of VRRP.

I'm afraid I only know the architectural design of our system; I can't
help you with specifics - I leave all the network bits to the network

You can of course do similar things with any number of commercial load
balancers, like an F5 or a BigIP.



To unsubscribe, e-mail:
For additional commands, e-mail:

View raw message