Arp Cache issue

[Georgios Magklaras]

On 07/13/2011 03:34 PM, brian irvin wrote:
> We are using bnx2 driver. We are getting outages when arp table fills up and unless we flush the table, network connectivity is an issue.
>
> more ifcfg-bond0
> DEVICE=bond0
> BONDING_OPTS="mode=1 miimon=500 primary=eth4"
>
>
> Thanks
>
> Brian

Assuming that you obviously update the system via RHN to have the latest 
and the greatest, I would check to see if your switch ports are 
configured in some sort of peculiar mode. If the rest of the boxes you 
run has a smaller arp table, then the switch port might be throwing a 
lot of traffic because, for example, it might be in SPAN mode, throwing 
all the traffic on the VLAN, when it does not need to be.

Failing that, if you have a large VLAN and more than a 1000 clients on 
it, depending on the class of your subnet, you can turn into the 
following ARP/IP kernel parameter:

sysctl -a | grep -i thresh

Have a look at this document:

http://www.clusterresources.com/torquedocs21/a.flargeclusters.shtml

which suggests amongst other things ARP flush parameters to put under 
/etc/sysctl.conf

/etc/sysctl.conf

# Don't allow the arp table to become bigger than this
net.ipv4.neigh.default.gc_thresh3 = 4096
# Tell the gc when to become aggressive with arp table cleaning.
# Adjust this based on size of the LAN.
net.ipv4.neigh.default.gc_thresh2 = 2048
# Adjust where the gc will leave arp table alone
net.ipv4.neigh.default.gc_thresh1 = 1024
# Adjust to arp table gc to clean-up more often
net.ipv4.neigh.default.gc_interval = 3600
# ARP cache entry timeout

net.ipv4.neigh.default.gc_stale_time = 3600


-- 
-- 
George Magklaras PhD
RHCE no: 805008309135525

Senior Systems Engineer/IT Manager
Biotek Center, University of Oslo
EMBnet TMPC Chair

http://folk.uio.no/georgios

Tel: +47 22840535