I'm running standard RH6.2, with pump-0.7.8-1. The machine has two ethernet cards, eth0 on the local network and eth1 connected to the Internet. eth0 is a static IP and eth1 is DHCP from my service provider. The machine has a 14 day uptime when the eth1 link was broken for several hours. During this time, the eth1 lease expired but when pump tried to renew it I got: Aug 12 10:10:16 dns1 pumpd[362]: failed to renew lease for device eth1 Aug 12 10:10:17 dns1 pumpd[362]: terminating as there are no more devices under management in /var/log/messages. So when the link was restored, I had to manually restart pump since it had exited. Pump has been associated with the eth1 interface via Linuxconf 1.19r2 by selecting "dhcp" for config mode on eth1.
You may have a look at bug 10348 for more information (probably the same bug).
I just got bitten by this bug (again) with pump-0.8.3-2, included with RedHat 7. In bug 10348, it was mentioned that the patch was applied. I haven't verified the source to see if that is true, but pump (after successfully obtaining a lease) will exit if any subsequent renewal fails (when it shouldn't).
Yes, pump-0.8.3-2 has been patched. However, as I mentioned more than 6 months ago in bug 10354, once bug 10348 is fixed, there is another bug that makes pump dies, if the network is broken while renewing the lease. These two reports were closed at the same time by Redhat when they fixed bug 10348. So they probably made a confusion. I have just reopened bug 10354.
Closed in conjunction with 10354
*** Bug 16928 has been marked as a duplicate of this bug. ***
*** Bug 18174 has been marked as a duplicate of this bug. ***