That kind of worked. I disabled KeepAlives in apache2 and now the boxes
don't get overloaded with 100 users, but it still does not load balance
watch -n1 "netstat -an | grep \":80 \" | wc -l"
yeilds ~20-30 on realserver 1 and ~1 on realserver2
pat
-----Original Message-----
From: lvs-users-bounces@xxxxxxxxxxxxxxxxxxxxxx
[mailto:lvs-users-bounces@xxxxxxxxxxxxxxxxxxxxxx] On Behalf Of Joseph
Mack
Sent: Tuesday, February 22, 2005 9:59 AM
To: LinuxVirtualServer.org users mailing list.; jmack@xxxxxxxx
Subject: Re: LVS+NAT not balancing properly
> I know that I am missing something here, but google won't yield the
> answers to me (after a few days of searching).
the search engine on the LVS documentation would be a better place to
look.
> In least-connection mode:
> About 98% of all connections hit realserver 1 with the remaining 2%
hitting realserver 2.
> In round-robin mode:
> 100% of all connections hit realserver 1 until it gets too busy and
goes down.
If the client is making the connections from the same IP _and_ the httpd
has persistence on (this is a different persistence to the ip_vs
persistenc), then none of your connections will close (look on the
realserver with `netstat` you'll only have one connection to the
client) and you'll always go to the same realserver.
Otherwise the connections should terminate (you'll have 100's of
FIN_WAIT on the realserver with netstat) and you should get
loadbalancing.
(Someone else was having trouble getting load balancing with http
requests
recently - but I never found out what the problem was)
Joe
--
Joseph Mack PhD, High Performance Computing & Scientific Visualization
LMIT, Supporting the EPA Research Triangle Park, NC 919-541-0007
Federal Contact - John B. Smith 919-541-1087 - smith.johnb@xxxxxxx
|