On Tue, 4 Mar 2008, Ganesh wrote:
> We are running a loadbalancer with ipvsadm-1.24-6. For the last few
> days we are getting a stiff increase in number of inactive connections
> (about 3800) in ipvs table. After observation I got that many connection
> entries which are either UDP or in TCP FIN_WAIT state keeps on resetting
> their TIMEOUT value to 60 sec. after it becomes zero.
> IPVS connection entries
> pro expire state source virtual
> destination
> UDP 00:59 UDP 220.227.223.25:54725 220.227.223.249:5000
> 97.242.123.4:5000
UDP is a difficult protocol to load balance - we haven't
figured out the best way to handle it - and you get odd
behaviours sometimes. All that we know about UDP and the
attempts at solutions is written up in the HOWTO - it's
scattered in a few spots. Have a look there to see if any of
it matches your problem
Joe
--
Joseph Mack NA3T EME(B,D), FM05lw North Carolina
jmack (at) wm7d (dot) net - azimuthal equidistant map
generator at http://www.wm7d.net/azproj.shtml
Homepage http://www.austintek.com/ It's GNU/Linux!
|