In Linux, is there a different arp cache timeout for non existent hosts?
At my router, 192.168.0.1, i'm seeing aprox. 8 new entries every 10 seconds for an IP that has been down for a long time.
Sample output from tcpdump -n -i eth1 broadcast:
18:54:21.107848 arp who-has 192.168.0.45 tell 192.168.0.1 18:54:21.607919 arp who-has 192.168.0.45 tell 192.168.0.1 18:54:22.107922 arp who-has 192.168.0.45 tell 192.168.0.1
gc_stale_time is set to 200:
# cat /proc/sys/net/ipv4/neigh/eth1/gc_stale_time 200
And yes, this ip is on my arp table:
arp -an | grep 192.168.0.45 ? (192.168.0.45) at <incomplete> on eth1
Why it keeps sending so many requests every 10 seconds?