ISC has discovered and fixed two memory leaks in the DHCP code. One of the leaks only affects servers running in DHCPv6 mode. The other is known to affect a server running in DHCPv6 mode but could potentially occur on servers running in DHCPv4 mode as well. In both cases the server can leak a small amount of memory while processing messages. The amount leaked per iteration is small and the leak will not cause problems in many cases. However on a server that is run for a long period without re-starting or a server handling an extraordinary amount of traffic from the clients the leak could consume all memory available to the DHCP server process, preventing further operation by the DHCP server process and potentially interfering with other services hosted on the same server hardware. Upstream has indicated that 3.1.x is potentially vulnerable to this flaw, although since it is no longer supported upstream, they have not investigated. They also indicate that these leaks are reproducable when running in DHCPv6 mode; one of the leaks only affects DHCPv6 mode, while the other may also theoretically affect DHCPv4 servers as well. A temporary workaround is to periodically restart the dhcp service.
This is now public: https://kb.isc.org/article/AA-00737
Created dhcp tracking bugs for this issue Affects: fedora-all [bug 842892]
dhcp-4.2.4-9.P1.fc17 has been pushed to the Fedora 17 stable repository. If problems still persist, please make note of it in this bug report.
This issue has been addressed in following products: Red Hat Enterprise Linux 6 Via RHSA-2012:1141 https://rhn.redhat.com/errata/RHSA-2012-1141.html
dhcp-4.2.3-11.P2.fc16 has been pushed to the Fedora 16 stable repository. If problems still persist, please make note of it in this bug report.