Services

I have 3 m0n0wall servers and I have had this happend to two diffrent servers in the last day.  It just stops handing out IP's the CPU load goes to 100% I look at a "top -d 1" and see that dhcpd is runnung 85% of the CPU.  When I check the DHCP log I see loads of these errors.

Mar 1 12:42:00 dhcpd: DHCPREQUEST for 172.19.65.247 (172.19.64.1) from 0c:77:1a:b9:80:c1 via 172.19.64.14: database update failed
Mar 1 12:42:00 dhcpd: Wrote 0 deleted host decls to leases file.
Mar 1 12:42:00 dhcpd: Wrote 0 new dynamic host decls to leases file.
Mar 1 12:42:00 dhcpd: write_lease: unable to write lease 172.19.66.49
Mar 1 12:42:00 dhcpd: DHCPREQUEST for 172.19.66.56 (172.19.64.1) from 10:9a:dd:d9:e0:3c via 172.19.64.12: database update failed
Mar 1 12:42:00 dhcpd: Wrote 0 deleted host decls to leases file.
Mar 1 12:42:00 dhcpd: Wrote 0 new dynamic host decls to leases file.
Mar 1 12:42:00 dhcpd: write_lease: unable to write lease 172.19.66.49
Mar 1 12:42:00 dhcpd: DHCPREQUEST for 172.19.64.246 (172.19.64.1) from 7c:c5:37:a1:50:c6 via 172.19.64.13: database update failed
Mar 1 12:42:00 dhcpd: Wrote 0 deleted host decls to leases file.
Mar 1 12:42:00 dhcpd: Wrote 0 new dynamic host decls to leases file.

Any Ideas as to what is causing this?   Also I did check my disk space and / was only 98% full.

The / partition was only 98% full? That's clearly indicative of a problem as most *nix systems will begin to flail or even crash before reaching that point.

Have you tried rebooting?

What hardware is m0n0 installed to and what version are you running?

The OS when it boots / is at 90%  I would go for more but the generic-pc image has that as the size of the drive. 

Also the version I am running is 1.3.3.

Rebooting fixes it at that point but this has happened two days in a row and I have a hunch that there is a setting somewhere that I have messed up that is causing the issue.

What about the memory, how much RAM is installed?

I have 4G of RAM in the system but that is not the issue.

I had the same issue today and same thing in the log but I looked at the system log and it states that the / file system is full. 

Strange thing is that when I did a "df" it is only at 97%.  I can only assume that something else is writing to / and combined with DHCP / gets filled.  I am going to lower my DHCP scope and see if that helps.

I don't think you understand the implications of having a file system "only" 97% full.

Most *nix OSs set aside around 5% of the available amount for use by the root user to be able to log in and clean up after a runaway process filled the file system up rendering the system unusable. The amount set aside is tunable, and can be disabled.

When usage gets beyond 95%, things go badly.

I don't know if m0n0wall has such a set aside in place, but FreeBSD, which is what m0n0wall is built on does. If this was not disabled, you are likely running into the point where things break.

I don't think m0n0wall can use any RAM beyond 128MB. And the file system is held in RAM disks. 

Your DHCP lease table maybe filling up the RAM disk.

Shorten the lease record time to live, that would get rid of unneeded lease entries.
Try with big shortening and then increase it as needed.