Problem: Device discovery fails because of exceeded ARP cache

ARP cache and /var/log/messages

When an NFM-P system manages a large number of NEs in a broadcast domain, the ARP cache on a main server station may fill and prevent the discovery of additional devices. When this happens, the /var/log/messages file contains entries like the following:

Jan 21 09:37:40 hostname kernel: Neighbour table overflow
Jan 21 09:37:40 hostname kernel: Neighbour table overflow
Jan 21 09:37:40 hostname kernel: Neighbour table overflow
Jan 21 09:38:00 hostname kernel: __ratelimit:190 callbacks suppressed

Perform this procedure when one of the following occurs:

The default ARP cache threshold values are the following:

Steps
 

Log in to the main server station as the root user.


Open a console window.


Perform one of the following to increase the ARP cache thresholds.

  1. To temporarily increase the thresholds, type the following:

    echo 8096 > /proc/sys/net/ipv4/neigh/default/gc_thresh1 ↵

    echo 25600 > /proc/sys/net/ipv4/neigh/default/gc_thresh2 ↵

    echo 32384 > /proc/sys/net/ipv4/neigh/default/gc_thresh3 ↵

  2. To permanently override the default thresholds, perform the following steps.

    1. Open the /etc/sysctl.conf file using a plain-text editor such as vi.

    2. Add the following lines to the end of the file:

      net.ipv4.neigh.default.gc_thresh1 = 8096

      net.ipv4.neigh.default.gc_thresh2 = 25600

      net.ipv4.neigh.default.gc_thresh3 = 32384

    3. Save and close the file.

    4. Enter the following:

      sysctl -p ↵


Close the console window.

End of steps