From: Chuck Church (cchurch@xxxxxxxxxxxx)
Date: Sat May 04 2002 - 14:18:38 GMT-3
Curtis,
MS uses what they call the 'dead router detection' algorithm, or
something like that. If it thinks the primary has failed, it will fail over
to the other. It's not a very intelligent algorithm, at least it wasn't in
'98. I had an NT 4.0 workstation that would just drop off the network
occasionally. Couldn't ping off his subnet. IPConfig showed the valid
default router. I was just starting to learn IP, so this is where I learned
to use the 'route' command on a machine. 'route print' showed 2 default
routes with different metrics. The correct one (the one shown in ipconfig)
was the top one with the lower metric. The second 0.0.0.0 route pointed to
an unknown next hop. I asked the server admin who controlled the DHCP
server, and he said he had put in the second default router, as our network
guys were installing it next week. What I think would happen is the user
would try to hit a web site that was either slow or not reachable. This
algorithm would kick in, not even checking that the original default gateway
was still reachable. Route print would still show the original router as
the first default, but the IP stack was at this point trying to use the
second one defined, which didn't exist yet. I'm hoping MS has fixed that in
later stacks, but I'm not sure. From now on, it's HSRP all the way! Sorry
to ramble...
Chuck Church
CCIE #8776, MCNE, MCSE
Sr. Network Engineer
Magnacom Technologies
140 N. Rt. 303
Valley Cottage, NY 10989
845-267-4000
-----Original Message-----
From: nobody@groupstudy.com [mailto:nobody@groupstudy.com]On Behalf Of
Curtis Phillips
Sent: Saturday, May 04, 2002 9:30 AM
To: ccielab@groupstudy.com
Subject: Behaviour of DHCP clients
Hello All,
I have been playing wiht Microsoft DHCP clients and wonder
based on my tests whether there is a benefit to an active DHCP client to
have
been sent multiple default-routers at DHCP lease time.
How does an active client become aware of the lack of availability of the
primary default? DNS seems understandable as the resolver is actively trying
to resolve the name and will process until resolved or out of options.
An end-system, on the other hand appears to send non-locally addressed
packets
blindly to a failed local default.
Any thoughts?
Your input is appreciated.
Curtis
This archive was generated by hypermail 2.1.4 : Thu Jun 13 2002 - 10:58:50 GMT-3