On Wed, Mar 12, 2008 at 10:12:19AM +0530, Indunil Jayasooriya alleged:

Hi,

I am getting below error on mailgw. it has 2 ethernets. eth0 is connected to internet, while eth1 is connected to LAN where there are about 300 PCs.

Mar 12 09:14:00 gateway kernel: NET: 697 messages suppressed. Mar 12 09:14:00 gateway kernel: Neighbour table overflow. Mar 12 09:14:05 gateway kernel: NET: 660 messages suppressed. Mar 12 09:14:05 gateway kernel: Neighbour table overflow. Mar 12 09:14:10 gateway kernel: NET: 682 messages suppressed. Mar 12 09:14:10 gateway kernel: Neighbour table overflow. Mar 12 09:14:15 gateway kernel: NET: 700 messages suppressed. Mar 12 09:14:15 gateway kernel: Neighbour table overflow.

Mar 12 09:14:20 gateway kernel: NET: 633 messages suppressed. Mar 12 09:14:20 gateway kernel: Neighbour table overflow.

it happens very often. I goolgled and found some info.

many talk about below file

/proc/sys/net/ipv4/neigh/default/gc_thresh1

I increased its value to 300

But, i still get the same error.

The "default" is only applied to devices as they come up. You'll need to also increase the values in the interface-specific directories.

Be sure to also adjust your /etc/sysctl.conf because manually playing in /proc won't survive a reboot.

In my very large flat network, I have the following in /etc/sysctl.conf: net.ipv4.neigh.default.gc_thresh1 = 4096 net.ipv4.neigh.default.gc_thresh2 = 8192 net.ipv4.neigh.default.gc_thresh3 = 8192 net.ipv4.neigh.default.base_reachable_time = 86400 net.ipv4.neigh.default.gc_stale_time = 86400

These are documented in arp(7).