Dhcpd: send_packet: No buffer space available

Tillebeck

I have this error in the log on boxes under heavy load. When it comes no new computers can get an IP but those that allready got an IP on LAN will still have access to the internet.
dhcpd: send_packet: No buffer space available

ROUTER 1:
On one router I see this entry once a day. And the router seems to work just fine. Is it anything to worry about?

ROUTER 2:
On another router I see the entry again and again. When it happens new clients cannot get an IP. A reboot will solve the issue, but it will happen again. Sometime there is a week in between. But once under constantly 25Mbit traffic between OPT1 and LAN it did happen three times over a few hours. The log is from one of those times.

Here is part of the log. It just goes on and on:

Aug 11 20:18:47

dhcpd: send_packet: No buffer space available

Aug 11 20:21:33

dhcpd: send_packet: No buffer space available

Aug 11 20:22:31

dhcpd: send_packet: No buffer space available

Aug 11 20:23:16

dhcpd: Abandoning IP address 192.168.24.146: declined.

Aug 11 20:23:29

dhcpd: Abandoning IP address 192.168.24.142: declined.

Aug 11 20:23:42

dhcpd: Abandoning IP address 192.168.24.175: declined.

Aug 11 20:23:53

dhcpd: send_packet: No buffer space available

Aug 11 20:23:56

dhcpd: Abandoning IP address 192.168.24.163: declined.

Aug 11 20:24:10

dhcpd: Abandoning IP address 192.168.24.134: declined.

Aug 11 20:24:23

dhcpd: Abandoning IP address 192.168.24.135: declined.

Aug 11 20:24:34

dhcpd: send_packet: No buffer space available

Aug 11 20:24:36

dhcpd: send_packet: No buffer space available

Aug 11 20:24:45

dhcpd: Abandoning IP address 192.168.24.147: declined.

Aug 11 20:24:58

dhcpd: Abandoning IP address 192.168.24.164: declined.

Aug 11 20:25:12

dhcpd: Abandoning IP address 192.168.24.172: declined.

Aug 11 20:25:25

dhcpd: Abandoning IP address 192.168.24.31: declined.

Aug 11 20:25:30

dhcpd: send_packet: No buffer space available

Aug 11 20:25:36

dhcpd: send_packet: No buffer space available

Aug 11 20:25:49

dhcpd: Abandoning IP address 192.168.24.20: declined.

Aug 11 20:26:02

dhcpd: Abandoning IP address 192.168.24.6: declined.

Aug 11 20:26:15

dhcpd: Abandoning IP address 192.168.24.114: declined.

Aug 11 20:26:28

dhcpd: Abandoning IP address 192.168.24.156: declined.

Aug 11 20:26:47

dhcpd: send_packet: No buffer space available

Aug 11 20:26:56

dhcpd: Abandoning IP address 192.168.24.145: declined.

Aug 11 20:27:09

dhcpd: Abandoning IP address 192.168.24.79: declined.

jimp

Are you maxing out your states? Or mbufs?

Anything else in the system log?

Tillebeck

Unfortunately I did not save any more logs. But I dit turn the pfsense of right after its last crash and it has not been turned on since. If the logs are not cleared on a reboot I should be able to boot it and pull the logs.

What should I looke for? Eg. the latest 200 log entries in system log…

The hardware vender has been really helpful and came with this suggestion:
"I have been looking into this problem and it seems to be an issue of a literally too small socket buffer.
You can increase the buffer size here (kern.ipc.maxsockbuf):"

This box has 2GB RAM and Maximum socket bufer size is set to: default (4262144)

There was less than 100 users on it during and probably less than 15 was active. It crashes when there is 25Mbit trafic (hard limit) between OPT1 and LAN. Not nearly heavy enough to bring this box down - I should think.

Thanks for taking your time.
BR. Anders

wallabybob

What is the output of pfSense shell command netstat -m

Perhaps you have had a temporary exhaustion of mbufs.