Iptables forwarding flapping

[AIM Systems]

I hope I can articulate my problem is as few words as needed.
I have two networks, each with their own firewalls and a IPsec connection between them
192.168.100.1 is one firewall(F8). It is also a virtual machine host for the web server: 192.168.100.5(F10)

192.168.700.1(F6) is my other firewall. 192.168.700.2(F6) is an application server behind it.

I have an OpenSwan 2.4.9 IPsec connection between these networks.

I have iptable forwarding rules that redirect any port 80 or 443 traffic directed at 192.168.100.1 towards 192.168.100.5 and source masquerading in the other direction

 
$IPT=/sbin/iptables

$IPT -A INPUT -s 0/0 -p tcp --dport 80 -j ACCEPT
$IPT -A FORWARD -p tcp -i $IF --dport 80 -o $INIF -d 192.168.100.5 -j ACCEPT
$IPT -t nat -A PREROUTING -p tcp -dport 80 -d 110.150.200.3 -j DNAT --to-destination 192.168.100.5:80

$IPT -A INPUT -s 0/0 -p tcp --dport 443 -j ACCEPT
$IPT -A FORWARD -p tcp -i $IF --dport 443 -o $INIF -d 192.168.100.5 -j ACCEPT
$IPT -t nat -A PREROUTING -p tcp -dport 443 -d 110.150.200.3 -j DNAT --to-destination 192.168.100.5:443

$IPT -t nat -A POSTROUTING -d 192.168.100.5 -o $INIF -j SNAT --to-source 192.168.100.1

Sporatically, a warning from my app server will say that 192.168.100.1 is down.
I'll ping 192.168.100.1 from my firewall (192.168.700.1) and all will be well
If I ping from the app server (192.168.700.2), no response happens.
When I do a:

traceroute 192.168.100.1

from 192.168.700.2
I get:

traceroute to 192.168.100.1 (192.168.100.1), 30 hops max, 60 byte packets
 1  wall.mysystem.com (192.168.700.1)  0.134 ms  0.061 ms  0.072 ms
 2  192.168.100.5 (192.168.100.5)  25.826 ms  26.041 ms  26.025 ms

The funny(strange) thing is if I perform a

service ipsec restart

This fixes things but only for a while.

Ditto with

service firestarter restart

Yet if I ignore the warning messages my server spews, the problem - whatever it is - fixes itself.

Any thoughts?

Specific additional info can be provided upon request.

Cheers