On Fri, 14 Jan 2011, in the Usenet newsgroup comp.os.linux.networking, in
article <5564c69d-6f36-4e20-b75a-(E-Mail Removed)>,
(E-Mail Removed) wrote:
NOTE: Posting from groups.google.com (or some web-forums) dramatically
reduces the chance of your post being seen. Find a real news server.
>Lew Pitcher <lpitc...@teksavvy.com> wrote:
>> rick.jon...@hp.com wrote:
>>> nc...@hoodcanal.com <hoodcanal...@usa.com> wrote:
>>>> With a decent sized arp cache I would assume arp requests from a
>>>> machine only appear every 15 min or so. But I am seeing the same
>>>> requests every 2 min or less.
What is in the arp cache? '/sbin/arp -a'
>>>> Is this unusual?
>>> Depends on the various stacks' settings for ARP cache aging.
and how often some application wants to see "who's out there"?
>> Which, for Linux 2.6 kernels, can be set by altering the contents
>> of various /proc/sys/net/ipv4/neigh/*/* files
>I am still a bit fuzzy on what is normal but I'll look into the
>proc/... files
0826 Ethernet Address Resolution Protocol: Or Converting Network
Protocol Addresses to 48.bit Ethernet Address for Transmission
on Ethernet Hardware. D. Plummer. November 1982. (Format:
TXT=21556 bytes) (Updated by RFC5227, RFC5494) (Also STD0037)
(Status: STANDARD)
1122 Requirements for Internet Hosts - Communication Layers. R.
Braden, Ed.. October 1989. (Format: TXT=295992 bytes) (Updates
RFC0793) (Updated by RFC1349, RFC4379, RFC5884) (Also STD0003)
(Status: STANDARD)
Use the search engine, and look for RFC1122 (and friends), then read
section 2.3.2. BRIEFLY - there is no "standard", but RFC1122 requires
the host to flush ``out-of-date'' ARP entries, and _suggests_ a timeout
on the order of a minute. It's a trade-off between having current
verses possibly wrong data and wasted space. If your hosts ALWAYS have
the same IP address, and different hosts are never assigned the same
address (one at a time) over long periods, then you can lengthen the
ARP timeout with little concern. On the other hand, an ARP request
and reply aren't taking up that much space on the wire, so there may
be little incentive to change the timeouts. RFC0826 also discusses
the rational.
Old guy