RE: Load balancing - Using only 2 servers-RH9.0

Edward Croft <[email protected]> 19 Aug 2003 10:35:39 -0400
Newsgroups gmane.linux.redhat.piranha
Message-ID <1061303738.19599.11.camel@crofts1>
On Tue, 2003-08-19 at 10:19, Sumesh wrote:
> Dear Edward,
> 
>    Right now I am looking for Load balancing of Web servers with 2
> Servers both acting as Director/Real Server.
>    This is different from my earlier requirement of Failover of Samba
> service.Sorry if I confused you.
>    In fact I have received a patch for doing this ( below mentioned)
> from Sebastien and I have applied this patch to
>    the piranha source given by you.
>  
>    Out of 2 servers Node1 and Node2 I have installed piranha in Node 1
> only.
>    Node 1 - IP address : 151.8.26.201
>    Node 2 - IP Address : 151.8.26.202
>    Node1 2nd NIC: 10.0.0.1(Real Server address)
>    Node 2 2nd NIC : 10.0.02(Real Address server)
>    Virtual Server: 151.8.26.210
>    NAT Router IP:10.0.0.254 
> 
>    Configured the default gateway as 10.0.0.254 for both Real Server
> IPs.
>    Edited /etc/httpd/conf/httpd.conf on both the servers as Server
> name  10.0.0.1 & 10.0.0.2 respectively.
>    Started httpd service on both the numbers.
> 
>    Started pulse on the Node1 and checked the ipvsadm -L and it has
> given a proper
>  
>  IP Virtual Server version 1.0.4 (size=65536)
>  Prot LocalAddress:Port Scheduler Flags
> 
>  RemoteAddress:Port Forward  Weight ActiveConn InActConn
> 
>  TCP 151.8.26.210:http wlc
>    -> 10.0.0.1:http         Masq     2     0              0
>    -> 10.0.0.2:http         Masq     1     0              0
> 
>   when I have checked from an outside client (http://151.8.26.210) it
> has given the page on Node1
>   But when I have stopped the httpd service from Node1 and tried above
> URL on client , it shown connection
>   refused and taking the page from Node2.
> 
>   Please let me know what actually went wrong and what this patch is
> doing.
> 
>  Thanks n Regards,
>  Sumesh
> 
The patch as far as I know only affects the way the backup server
monitors the primary in failover. The default piranha distribution only
allowed for monitoring http in failover, with no scripting. Sebastiens
patch allows scripting in failover as well as scripting all ready in
load balancing. 
Now as to the second part, you said "you stopped the service on Node1
and it showed connection refused, and taking the page from Node2?" Did
it get the page from node2? Connect refused may not have anything to do
with piranha, it could be a misconfigured apache. Have you tried
bringing that box up directly by pointing your browser to 10.0.0.2?