Problem with Direct Routing and Ultramonkey

[Brian Sheets]

I have 3 real servers running postfix and one server running UM in a DR config

checktimeout=60
checkinterval=30
autoreload=no
logfile="/var/log/ldirectord.log"
#logfile="local0"
quiescent=yes

virtual=172.16.11.47:25
        real=172.16.11.44:25 gate
        real=172.16.11.45:25 gate
        real=172.16.11.46:25 gate
#       real=172.16.11.48:25 gate
#       real=172.16.11.50:25 gate
        protocol=tcp
        service=smtp
        checktype=connect
        scheduler=wlc

IP Virtual Server version 1.2.0 (size=4096)
Prot LocalAddress:Port Scheduler Flags
  -> RemoteAddress:Port           Forward Weight ActiveConn InActConn
TCP  172.16.11.47:25 wlc
  -> 172.16.11.44:25              Route   1      0          0
  -> 172.16.11.45:25              Route   1      0          0
  -> 172.16.11.46:25              Route   1      0          0


I can make a connection to the VIP on port 25, and sometimes the connection lasts for a whole smtp chat, other times it doesn't, and then the connection is dropped and subsequent connections are refused.

I can still telnet to the real server port 25's and an ipvsadm -L -n yields the above table.

the only way I can get the service working again is to stop and start heartbeat and then of course it only works for one or part of one connection

Can anyone help

Brian