Massive "lease imbalance - lts" then silent crash

[ashley.hatch at unlv.edu]

We have been using ISC DHCPd with failover for about 5 years and in the 
only in the last 9 months have been running into a problem with the 
failover servers. 
Basically at a random point which may be days or weeks between one of the 
two server, seemingly always now the primary, will stop responding to DHCP 
requests and spews "lease imbalance - lts" into the logfiles by the 
hundreds. I am used to the LTS messages, as they are part of the standard 
log, but this is uninterrupted 30 messages/s until the daemon either:
A) crashes with no log entry or core file OR 
B) eventually recovers after 1-10min of not answering DHCP.

Log entries look like:
Jun  5 15:39:13 merry dhcpd: lease imbalance - lts = 2
Jun  5 15:39:13 merry dhcpd: lease imbalance - lts = 10
Jun  5 15:39:13 merry dhcpd: lease imbalance - lts = 7
Jun  5 15:39:13 merry dhcpd: lease imbalance - lts = 11
Jun  5 15:39:13 merry dhcpd: lease imbalance - lts = 10
Jun  5 15:39:13 merry dhcpd: lease imbalance - lts = 2
Jun  5 15:39:13 merry dhcpd: lease imbalance - lts = 10
Jun  5 15:39:13 merry dhcpd: lease imbalance - lts = 14
Jun  5 15:39:13 merry dhcpd: lease imbalance - lts = 6
Jun  5 15:39:13 merry dhcpd: lease imbalance - lts = 2

I am currently running version 3.0.4-13  on Debian 3.1. I have tried 
running hardware diagnostics on the boxes with no problems found and I 
have captured traffic during a "timeout" and traffic can flow into and out 
of the box, but it refuses to reply to DHCP requests.

Any help or direction would be appreciated. We did not seem to have this 
problem until we started using redundant HSRP routers to act as DHCP 
helpers per subnet.