[Packetfence-users] dhcpd stops responding to requests / iptalbes problem?

[Shelton, B. <bsh...@iw...>]

I am using packetfence in "dhcp mode"  ( Centos 5 with packetfence installed via yum (RPMs) and all packages at current versions - except for perl-Parse-RecDescent )

I have an issue where frequently (it seems randomly) dhcpd stops responding to requests.

Issuing "pfcmd service dhcpd restart" gets the system responding to requests again.  I've added that as a cron task that runs every half hour and that has helped considerably, but occasionally I need it to run every 15 minutes. ( it is on a system that handles around 1000 clients per day but I'm not seeing any CPU, I/O, or memory bottlenecks)

I did notice ONE error:
[Fri Jan 29 12:03:55 2010] [error] [client 172.31.178.246] iptables-restore v1.3.5: Unknown arg `--set-mark'
[Fri Jan 29 12:03:55 2010] [error] [client 172.31.178.246] Error occurred at line: 1620
[Fri Jan 29 12:03:55 2010] [error] [client 172.31.178.246] Try `iptables-restore -h' or 'iptables-restore --help' for more information.
[Fri Jan 29 12:04:13 2010] [error] [client 172.31.178.246] iptables-restore: COMMIT expected at line 2694
[Fri Jan 29 12:04:15 2010] [error] [client 172.31.178.246] iptables-restore: COMMIT expected at line 549

But it only occurred once and the "dhcp stopped responding to requests" problem is considerably more frequent than that.

I did update perl-Parse-RecDescent at one point and had to back it off as well as rebuilding pfcmd_pregrammar.pm.  But other than that I haven't noticed any other issues with the box.

I'm thinking it is most likely iptables blocking the dhcp requests (since "pfcmd service dhcpd restart" "restarts" iptables too)


Has anyone else out there experienced anything similar?

Does anyone know if there are any problems with "restarting" dhcpd a lot?  i.e. any  possibilities where something might step on something because dhcp was restarting?

Thanks