Re: [Netdisco] MAC addresses not sticking on switch

[Brandon M. <Bra...@wr...>]

Looks like deleting both devices and discovering it with the main 172.16 =
IP address worked.  The device is a Cisco 3550 48 port switch.  Thank you =
for all your help Max!!

Brandon

>>> Max Baker <ma...@wa...> 11/22/05 04:18PM >>>
Weirdness.

Looks like you have two devices in netdisco for this one device. =20

Try removing both and rediscovering the main interface first.  What kind
of device is this btw?=20

You should see one entry in the 'device' table that is for 172.16.4.254
and one entry in the device_ip table that is :

>  172.16.4.254 | 192.0.0.239  | GigabitEthernet0/2  |     | 2005-11-22 =
09:00:31.296958

My guess is that the device is answering SNMP calls on 192.0.0.239 but
when doing so it decides that 192.x is the main IP instead of 172.x.  The
converse is happening when answering SNMP on 172.x.  This is probably a
corner case netdisco hasn't seen before and needs something in there to
make it decide that 172.16 is the main IP when answering calls on =
192.X.=20

For now try just removing both devices and discovering the "main" one
first.

-m

On Tue, Nov 22, 2005 at 04:05:59PM -0500, Brandon Molnar wrote:
> Hello again Max,
>=20
> It looks like all those MAC addresses are being archived when each =
macsuck cronjob runs.  I'm starting to see a pattern though.  This switch =
has 2 IP addresses, 172.16.4.254 and 192.0.0.239  The MAC addresses that =
"stick" (like they should) show the MAC being on 172.16.4.254 while all =
the archived ones show 192.0.0.239.  When I manually run a macsuck on =
172.16.4.254 all the MAC addresses are unarchived.  If I run a manual =
macsuck on 192.0.0.239, they go back to archived.
>=20
> In in device_ip table I see:
>       ip      |    alias     |         port        | dns |          =
creation
> --------------+--------------+---------------------+-----+---------------=
-------------
>  172.16.4.254 | 192.0.0.239  | GigabitEthernet0/2  |     | 2005-11-22 =
09:00:31.296958
>  192.0.0.239  | 172.16.4.254 | Vlan1               |     | 2005-11-22 =
09:00:40.198045
>=20
> So to fix this, would I need to change something in Netdisco or on the =
switch itself?
>=20
> Brandon
>=20
> >>> Max Baker <ma...@wa...> 11/22/05 02:34PM >>>
> Hi Brandon,
>=20
> MAC addresses won't just disappear unless you have the expire settings =
in
> netdisco.conf set to something extremly frequent.  MAC addresses can =
move
> to other places though if they are incorrectly seen on other ports that
> are not marked as uplinks.   Search on one of the MAC addresses in
> question and see where else it might live.  Also try the archive option =
to
> see all the places that it's been picked up at.=20
>=20
> -m
> On Tue, Nov 22, 2005 at 02:18:43PM -0500, Brandon Molnar wrote:
> > Thank you for your reply Alan.  I checked the MAC address table =
timeout on that switch and it's set to 5 minutes like all my other =
switches.  The addresses from the switch should only expire if it doesn't =
see any traffic from that address within that 5 minute span.
> >=20
> > I think it's something in Netdisco because if I manually run a =
macsuck, all but 1 or 2 of the ports will usually show a MAC address =
connected to it.  Just after the macsuck cron job completes, most of them =
disappear.  If I manually run a macsuck again 1 minute later, they all =
come back.
> >=20
> > The only major thing I can see that is different between this switch =
and all my others is that this one has 2 IP addresses.  Could that somehow =
cause my problem?
> >=20
> > Brandon
> >=20
> > >>> <A.L...@lb...> 11/22/05 04:05AM >>>
> > Hi,
> >=20
> > > Any ideas?
> >=20
> > read IOS documentation, iirc the default aging time for MAC addresses =
on
> > such a device is 5 minutes. you'll want to think about that...and =
there are
> > pros and cons to changing this...heres your free starter  :-)
> >=20
> > the default setting of the "mac-address-table aging-time" global =
configuration command
> >=20
> > alan
> >=20