Memory Leak in dhcpd

[Mathew, Meera]

Hello,
I have been running ISC dhcp version 3.1.2 in failover mode including a patch one-lease-per-mac on Linux RedHat Server 5.1.

The configuration file has only 3 pools configured to it with 20000 ips each (total pool size of 60000). Lease time is 10 hours.

A performance test was done for the failover pair, where the scenario was that both servers in failover pair are up and running normally, but the 60000 client requests are sent only to the Primary server (simulating possible router malfunction)
The Primary server hands out 30000 leases in the first hour, But after an hour the %memory consumption of the dhcpd process significantly increases from about 8% to 90% and finally leads to the dhcp server to crash with the following messages at the end of the 2nd hour.


no memory for option buffer.
no memory for option buffer.
no memory for option buffer.
dhcp_failover_put_message: something went wrong.
dhcp_failover_send_connect: out of memory
Segmentation fault

The same thing happens if the secondary is the only server receiving requests.
I have searched the mailing lists and found people have faced similar issues but have not got a solution.
Does any body have any idea why this happens?
Any help will be appreciated.

Thanks in advance
Meera.
</PRE><p style="font-family:arial;color:grey" style="font-size:13px">This message contains information that may be privileged or confidential and is the property of the Capgemini Group. It is intended only for the person to whom it is addressed. If you are not the intended recipient, you are not authorized to read, print, retain, copy, disseminate, distribute, or use this message or any part thereof. If you receive this message in error, please notify the sender immediately and delete all copies of this message.</p><PRE>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.isc.org/pipermail/dhcp-users/attachments/20090813/0dc976cc/attachment.html>