VPN 3030 - VPN Client 4.x - loss of internal network access after 4 hours

Ever since I upgraded the VPN code to the latest release
(vpn3000-4.7.1.Rel-k9.bin) users are now losing connection to the
internal network at the 4 hour mark. They stay connected to the VPN,
but can only ping the external interface of the concentrator. Session
timeouts are set at 48 hours.

Here is the log of the client, 13:00 is the 4 hour mark.

659 12:59:56.493 07/11/05 Sev=Info/6 IKE/0x63000054
Sent a keepailve on the IPSec SA

660 13:00:04.005 07/11/05 Sev=Info/6 IKE/0x6300003D
Sending DPD request to x.x.x.x, seq# = 3874843677

661 13:00:04.005 07/11/05 Sev=Info/4 IKE/0x63000013
SENDING >>> ISAKMP OAK INFO *(HASH, NOTIFYPD_REQUEST) to x.x.x.x

662 13:00:04.065 07/11/05 Sev=Info/5 IKE/0x6300002F
Received ISAKMP packet: peer = x.x.x.x

663 13:00:04.065 07/11/05 Sev=Info/4 IKE/0x63000014
RECEIVING <<< ISAKMP OAK INFO *(HASH, NOTIFYPD_ACK) from x.x.x.x

664 13:00:04.065 07/11/05 Sev=Info/5 IKE/0x6300003F
Received DPD ACK from x.x.x.x, seq# received = 3874843678, seq#
expected = 3874843678

665 13:00:14.521 07/11/05 Sev=Info/6 IKE/0x6300003D
Sending DPD request to x.x.x.x, seq# = 3874843678

666 13:00:14.521 07/11/05 Sev=Info/4 IKE/0x63000013
SENDING >>> ISAKMP OAK INFO *(HASH, NOTIFYPD_REQUEST) to x.x.x.x

667 13:00:14.571 07/11/05 Sev=Info/5 IKE/0x6300002F
Received ISAKMP packet: peer = x.x.x.x

668 13:00:14.571 07/11/05 Sev=Info/4 IKE/0x63000014
RECEIVING <<< ISAKMP OAK INFO *(HASH, NOTIFYPD_ACK) from x.x.x.x

669 13:00:14.571 07/11/05 Sev=Info/5 IKE/0x6300003F
Received DPD ACK from x.x.x.x, seq# received = 3874843679, seq#
expected = 3874843679

670 13:00:25.037 07/11/05 Sev=Info/6 IKE/0x6300003D
Sending DPD request to x.x.x.x, seq# = 3874843679

671 13:00:25.037 07/11/05 Sev=Info/4 IKE/0x63000013
SENDING >>> ISAKMP OAK INFO *(HASH, NOTIFYPD_REQUEST) to x.x.x.x

672 13:00:25.097 07/11/05 Sev=Info/5 IKE/0x6300002F
Received ISAKMP packet: peer = x.x.x.x

673 13:00:25.097 07/11/05 Sev=Info/4 IKE/0x63000014
RECEIVING <<< ISAKMP OAK INFO *(HASH, NOTIFYPD_ACK) from x.x.x.x

674 13:00:25.097 07/11/05 Sev=Info/5 IKE/0x6300003F
Received DPD ACK from x.x.x.x, seq# received = 3874843680, seq#
expected = 3874843680

675 13:00:35.553 07/11/05 Sev=Info/6 IKE/0x6300003D
Sending DPD request to x.x.x.x, seq# = 3874843680

676 13:00:35.553 07/11/05 Sev=Info/4 IKE/0x63000013
SENDING >>> ISAKMP OAK INFO *(HASH, NOTIFYPD_REQUEST) to x.x.x.x

677 13:00:35.614 07/11/05 Sev=Info/5 IKE/0x6300002F
Received ISAKMP packet: peer = x.x.x.x

678 13:00:35.614 07/11/05 Sev=Info/4 IKE/0x63000014
RECEIVING <<< ISAKMP OAK INFO *(HASH, NOTIFYPD_ACK) from x.x.x.x

679 13:00:35.614 07/11/05 Sev=Info/5 IKE/0x6300003F
Received DPD ACK from x.x.x.x, seq# received = 3874843681, seq#
expected = 3874843681


This DPD (Dead Peer Detection) transaction occurs once at the 3 hour
mark, but at the 4 hour mark it occurs over and over. I've seen one
related case similar to this issue, but it was VPN on a PIX, not a
concentrator.

Thanks for any help!

Nick