LVS
lvs-users
Google
 
Web LinuxVirtualServer.org

Re: why LVS-DR rr mode not average?

To: "LinuxVirtualServer.org users mailing list." <lvs-users@xxxxxxxxxxxxxxxxxxxxxx>
Subject: Re: why LVS-DR rr mode not average?
From: Joseph Mack <mack.joseph@xxxxxxx>
Date: Wed, 07 Jan 2004 06:17:11 -0500
lijixin wrote:
> 

> 3 hours after running situation:
> 
> IP Virtual Server version 1.0.9 (size=65536)
> Prot LocalAddress:Port Scheduler Flags
>   -> RemoteAddress:Port           Forward Weight ActiveConn InActConn
> TCP  192.168.1.214:http rr
>   -> 192.168.1.116:http           Route   1      1          410
>   -> 192.168.1.223:http           Route   1      85         177
> 
> who could tell me why??
> thanks a lot!!

this is something that we've known about for a while. See

http://www.austintek.com/LVS/LVS-HOWTO/HOWTO/LVS-HOWTO.ipvsadm.html#reusing_ports

The explanation is that some clients are re-using ports and that the director's
estimate of InActConn is based on assumptions about the behaviour of the 
connection between the client and the realservers. A better estimate
of the balance would be to see the number of ESTABLISHED and TIME_WAIT 
connections at the realservers, but I didn't think of that at the time. 

We haven't actually tested this theory with different types of clients 
(ones that re-use ports and ones that don't), so we don't know for sure.

We aren't really regarding this as a big enough problem to find a fix for.

Joe


-- 
Joseph Mack PhD, High Performance Computing & Scientific Visualization
SAIC, Supporting the EPA Research Triangle Park, NC 919-541-0007
Federal Contact - John B. Smith 919-541-1087 - smith.johnb@xxxxxxx
<Prev in Thread] Current Thread [Next in Thread>