Dennins,
Thanks for your input. I'll try to answer your questions inline:
"Dennis Bailey" wrote in message
[EMAIL PROTECTED]">news:[EMAIL PROTECTED]...
> How many remote sites off each 7500?
Anywhere from 15 to 30 remote sites off of each 7500 (via ATM subinterfaces)
> Are the remote sites all 'stub' sites?
As far as I can verify, each remote site is totally stubby (i.e. 1 router
in/out). However, we're not using OSPF so I can't make it an "official"
totally stubby area =) (BTW, that was a joke. I realize your usage of
'stub' simply meant it had no other 'backdoor' connections to our network)
> Do the remotes have connections to both 7500's (ie. primary and backup
pvc)?
I do not believe that any of the remote sites connect to both 7500s. I will
check on that, but for now let's assume they do not.
> The passive-interface idea you have is probably a good one.
Since both 7500s connect to their respective 5500 via a switchport in VLAN
1, I assume I can implement passive interface on all other VLANs that the
5500's RSMs see. This gets to the heart of my issue tho. Does having more
than one connection (whether physical cables or logical VLANs) between two
routers (RSMs) screw up EIGRP? Shouldn't it still only see 1 neighbor (that
it has multiple links to)?
> If your remote sites have a connection to both 7500's, make sure that you
> have outbound distribution lists on the remote routers so that the remotes
> don't advertise anything but what is behind them.
Good call. Again, I am 99% sure that each remote site is a totall stubby
network.
> If your addressing is organized in such a way that you can summarize
> everything behind each of the 7500's back into the core RSM's that would
> probably also help. This way if a remote link bounces, the 7500 will see
> the change and deal with it but the RSM's will not be affected.
Luckily the addressing is setup in such a way that I can summarize. I was
going to even try to one better that by summarizing at the remote router
(i.e. so that any changes on that LAN wouldn't send EIGRP updates over the
WAN to the 7500). However (get this), the routers on the remote side of the
WAN links are "controlled" by each individual location. I only have control
over the "centralized" 7500s and 5500s in the core (love those politics).
So in an effort to minimize EIGRP updates in the core, I have introduced a
summary route for the site that if the primary cause of these EIGRP updates
that start the storms. I configured the EIGRP summary route on the
FastEthernet interface that connects the 7500 to the 5500. (i.e. the remote
site EIGRP updates still come to the 7500, but then the 7500 summarizes on
it's outgoing interface to the 5500 as you mention above). So, at least for
this one remote site, when changes happen, the 7500 gets the update, but
everything else in the network (except the other WAN links hanging off of
that same 7500) are shielded from the updates because of the summary.
I didn't get a chance to capture a sniff of the recent storm (today), but
even since I introduced this summary address (actually, I introduced 2 of
them, one for each of the 2 main offending WAN sites) we've had these "EIGRP
storms" and shoot CPU usage up..... luckily for us it's not enough to make
them go to 80-90% or lock up the box, but it's enough to concern me. I'm
going to capture all day tomorrow so if there's another storm, I can see
which route is being stormed between the 5500s and try to "shut it down"
(i.e. introduce a summary in the appropriate place).... we'll see
Any further comments or input is MORE than welcome =)
Mike W.
> Dennis
>
> ""Michael L. Williams"" wrote in message
> [EMAIL PROTECTED]">news:[EMAIL PROTECTED]...
> > Hey all...... kind of a silly question, but how does EIGRP handle
updates
> > when there are multiple links between two routers......?
> >
> > Here is the scenario I've got:
> >
> > I have two 5500s connected via etherchannel trunks and a 7500 hanging
off
> > each one for WAN connections like so
> >
> > WAN -- 7500 --FastEther--5500 --Etherchannel Trunk -- 5500 --
FastEther --
> > 7500 -- WAN
> >
> > (careful of wrap)
> > FYI: The 7500s connect to their respective 5500 via FastEther that is
in
> > VLAN 1 (no flames!)
> >
> > Every once in a while an EIGRP update comes in from the WAN, and all of
a
> > sudden the two 5500's get "hung" in an EIGRP storm... I did a sniffer
> > capture (only in the VLAN I was connected to) and there were over 4,000
> > EIGRP updates from the two 5500's in 6 seconds..... (all for the same
> route
> > that was passed to them from the WAN by one of the 7500's). CPU goes
from
> > ~5% to 30% when this happens and the only way to stop it is to "clear ip
> > eigrp neighbors". Once this command is issued, it's all over.....
things
> go
> > back to normal (until it happens again)......
> >
> > My question is this. Since both RSMs on the 5500s see the same VLAN
> > interfaces (since all of the VLANS are trunked between the two
switches),
> > they effectivly have 12 (# of VLANs) connections to each other. Could
> this
> > be causing this "storm" of EIGRP updates?
> >
> > I was under the impression that the RSMs would see each other as a
single
> > EIGRP neighbor. I've considered taking all VLAN interfaces (in both
> RSMs)
> > except VLAN 1 (which connect to the 7500s) and making them passive
> > interfaces.....
> >
> > Any input here? I don't have a *clue* what could be causing this EIGRP
> > storm...... I can't say 100%, but I don't think the two 7500s are
> affected
> > by this storm.....
> >
> > TIA,
> >
> > Mike W.
Message Posted at:
http://www.groupstudy.com/form/read.php?f=7&i=14598&t=14579
--------------------------------------------------
FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html
Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
