dhclient appears to not handle protocol timeouts correctly
-
I ran into an issue with pfSense 2.4 where it seemed like some sequence or timing of events with regards to the WAN link state and the upstream link state of the attached modem would cause a persistent loss of the DHCP assigned WAN IP.
After investigating such behavior, I have come to the conclusion that pfSense mishandles DHCP timeouts, resulting in a case where if a DHCP protocol timeout occurs and the default gateway obtained in the last cached lease is not pingable, the DHCP protocol will not restart until the cached lease expires, as opposed to after the defined "retry" interval.
I have provided a more low level analysis of my observations in bug https://redmine.pfsense.org/issues/9267
Does anyone else see this behavior and/or agree with this analysis?