ARP or Spanning-tree problem - 4506 with 2950's

Unanswered Question
Jul 13th, 2004
User Badges:

I would like to clear out all ARP and clean all Spanning-tree info from all switches. I have a core switch (4506) with 2950's connected with either fiber or copper. I would like to perform a rediscover of all switches and devices connected to all switches.

Due to construction, I had to physically move a server room (used the same switch and physical cable, which trunked back to the 4506) and when I brought the switch (2950) back online I have been having intermittent connectivity issues to the servers that are connected to that same 2950. Some services work (DNS, etc.) others do not (DHCP, Relay Agents, etc.).

thanks for any help,


  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 0 (0 ratings)

I would start with a "clear counters" and see if you have any lower level issues first. DHCP and DHCP relay might be suffering from spanning tree startup delay. Do you have portfast enabled on all ports that are not connected to other switches ? If not you might want to set this up. It will reduce the spanning tree startup delay significantly. You can enable this on the 4500 or 2950 globally for all access ports by "spanning-tree portfast default" or per interface by "spanning-tree portfast".

Issue a "clear mac-address-table dynamic" on the 4506 to clear the cam table.

Issue a "clear mac address-table dynamic" on the 2950 to clear the cam table.

CiscoATC Tue, 07/13/2004 - 08:48
User Badges:

I have cleared the tables using the above commands. Still having an issue. I have an VPN Server (windows) and an FTP server (windows) where I am having the most difficulty connecting.

Before the move, the servers connected to this switch (2950 - hasn't changed, still the original switch) worked fine. I powered down the servers, the 2950 and then made the physical move to the new room. When I brought it all back up, its as if the network isn't seeing these servers.

Still plugging away at it. Any more suggestions?

thanks in advance,


steve.busby Tue, 07/13/2004 - 09:57
User Badges:
  • Silver, 250 points or more

I'd suggest going back to the basics.

- Speed/duplex: verify both sides are locked in at 100/full

- Cabling: check to ensure it's within specs & verify via MAC address that each server is connected where you think it is.

- Verify port configurations on the 2950 didn't get corrupted, can you verify it against a saved copy from before the move?

CiscoATC Tue, 07/13/2004 - 10:56
User Badges:

- speed and duplex are ok for each port (both sides)

- cables are factory; I had them tested again.

- cables were labeled as to the port that each server was connected; following the move cables were reinserted to the exact same port.

- I have verified the configs from recently saved copies. No changes have been made.

- Also - VTP data is consistent, same domain, all 2950's are set to client mode, while the 4506 is set to server mode

The only apparent variable was the physical move, which has led me to believe that I am having layer 2 problems (either VTP, Spanning-tree, ARP, etc.); possibly a rediscover issue or "bad" table somewhere.

CiscoATC Tue, 07/13/2004 - 16:33
User Badges:

An interesting note:

- if I change the NIC Card in one of the servers, I am able to have services working again across all VLANS and external access to the FTP Server.

This seems to lead toward a MAC address problem or some kind of layer two issue.

Any suggestions as to how to get the switches to rebuild all mac tables on all switches?

steve.busby Wed, 07/14/2004 - 10:08
User Badges:
  • Silver, 250 points or more

clear mac-address-table dynamic

or simply reboot the switch(es).

Check your mac-address-table aging-time to see how often mac-addresses are dropped.

sho mac-address-table aging-time

You say you "change the NIC Card" in one of the servers, is that to say you've swapped one card for another and now everything works fine? If so, then you've narrowed down your problem to a faulty NIC.

CiscoATC Thu, 07/15/2004 - 02:53
User Badges:

I would have thought that for the first server, but after changing two other NIC cards on two other servers (all connected to the same switch, 2950) I was really beginning to wonder that maybe I have a bad ARP/MAC table(s) being spread around my switches.

I have a visio of the network if you would like to take a look at the physical topology. I will "highlight" the 2950 which seems to be causing the issue. Just let me know if the drawing would help.


This Discussion