dhcpd crashing and and an "out of memory" being reported

[Oscar Ricardo Silva]

We're running V3.1.0 on two servers in a failover configuration. 
Earlier today one of the servers, DHCP1, crashed and the operating 
system reported an "out of memory" error.  The second of the two servers 
took over and no outage was experienced.  This is the third time we've 
seen this error this year, the first time in April, the second two weeks 
ago and then today.

The server in question is running Red Hat Enterprise Linux Server 
release 5.1, kernel 2.6.18-53.1.4.el5PAE and it has 8GB of memory.  I 
don't see evidence of any DOS/DDOS or an excessive number of 
DHCPDISCOVER (or other DHCP related requests).  Looking at  sar  output 
there's plenty of CPU available (99% idle), 2GB of memory available and 
network traffic is OK.  We did notice that after the dhcpd process died, 
free memory went from 2GB to 5GB.

Are there known memory leaks?  memory limits?  I saw something in 
passing about an OMAPI related leak.  We do use OMAPI to check the 
failover status and to restart the server after a configuration change.



Oscar