(PM) arp cache clearing/timeout on PM2

Jason Campbell (jdc_pmusers@introrse.com)
Tue, 25 May 1999 15:32:21 -0700 (PDT)

I'm having trouble making a rudimentary web server failover solution
work because the arp cache on our PM 2e (ComOS 3.1.3) takes *so* long
to timeout. When I simulate a failure of our primary server, the
second server aliases the first server's IP properly and all of our
local workstations can access it just fine, but the Portmaster doesn't
re-arp for the failed server for nearly an hour. Of course, during
that time all of our remote users are cut off completely, even though
the second server has already stepped in.

Nothing I've tried short of power-cycling the Portmaster seems to
cause it to flush that arp cache entry. (Including the obvious and
usual pinging of the PM from the first-server's IP on the second
server.) Searching the manual, the pm-users archive, and the web at
large hasn't yielded any solutions either.

I'd appreciate anything anyone can suggest. Ideally I'd like to
either drastically shorten the cache timeout (to 5 minutes or so), or
find a way to have the second server flush the Portmaster's arp cache
once it goes into failover mode.

Clearly we're quite far backlevel on ComOS, and we ought to think
about upgrading that, but I'm reticent to just blindly change something
that a) works now, and b) is so integral to our Internet connectivity
without any indication that the change will help. So if anyone has an
idea as to whether 3.7.? might fix this, that would help too.

Thanks very much, in advance,

Jason Campbell
-
To unsubscribe, email 'majordomo@livingston.com' with
'unsubscribe portmaster-users' in the body of the message.
Searchable list archive: <URL:http://www.livingston.com/Tech/archive/>