DHCP not giving out IP in less than 24hrs
-
im on the latest nanobsd snapshot, most of the issues r solved regarding pppoe etc but just noticed this DHCP issues yesterday, pfsense was on for less than 24hrs and then clients started reporting that they r not getting IP from the DHCP server, keeps saying limited connectivity so i guess this issue still remains.
earlier snapshots used to show some error related to DHCP during boot in the console but this newer built doesn't show any such error then y does it die in under 24hrs?
-
What's in the system logs? Is dhcpd actually running?
-
actually when it happens, then i cant connect to pfsense even using a fixed ip, only if a client is connected before the dhcp fail then that client has access to pfsense and the internet and system log does show anything nor does the dhcp log
-
Sounds like you're running out of states possibly?
-
y would running out of states make DHCP stop assigning IP addresses?
bytheway, even with no clients connected and just one client trying to get a DHCP lease, pfsense wont give it, making it totally useless untill a reboot.
-
on dhcp restart i get this in system log, dont know if its normal or something missing in it
Apr 2 12:32:56 dhcpd: Internet Systems Consortium DHCP Server V3.0.7 Apr 2 12:32:56 dhcpd: Copyright 2004-2008 Internet Systems Consortium. Apr 2 12:32:56 dhcpd: All rights reserved. Apr 2 12:32:56 dhcpd: For info, please visit http://www.isc.org/sw/dhcp/ Apr 2 12:32:56 dhcpd: Wrote 0 deleted host decls to leases file. Apr 2 12:32:56 dhcpd: Wrote 0 new dynamic host decls to leases file. Apr 2 12:32:56 dhcpd: Wrote 1 leases to leases file. Apr 2 12:32:56 dhcpd: Listening on BPF/vr0/00:0d:b9:13:47:84/192.168.0/24 Apr 2 12:32:56 dhcpd: Sending on BPF/vr0/00:0d:b9:13:47:84/192.168.0/24 Apr 2 12:32:56 dhcpd: Sending on Socket/fallback/fallback-net
-
y would running out of states make DHCP stop assigning IP addresses?
Because if you have no states, it can't receive the DHCP request, nor send the response. That's also consistent with the other behavior mentioned. Check your RRD graphs to see if that's the case.
-
my state table is set for 1000 and at max 800 entries appear so i doubt it being filled and regarding RRD graphs, i have them disabled
-
1000 is very, very small, I'm sure that's what you're hitting. In almost all cases you should never run with less than 10,000 unless you want to start dropping connections. If you see 800 from time to time, it doesn't take much at all to bump that up to 1000.
-
Given the importance of the state table size, it could be worthwhile adding sysctls to report the maximum table size and count of times the table was full and report these through the web GUI.