6509 Switchports are up, but client machine not passing traffic

wilson_1234_2 · ‎01-03-2008

We have a 6509 12.2(18)SXF5 with 720 Supervisor Engine,

and these modules:

1 5 Communication Media Module

2 8 8 port 1000mb GBIC Enhanced QoS

3 48 48 port 10/100 mb RJ45

4 48 48 port 10/100 mb RJ45

5 48 48 port 10/100 mb RJ45

6 2 Supervisor Engine 720 (Active)

7 48 48 port 10/100 mb RJ45

8 48 SFM-capable 48 port 10/100/1000mb RJ45

9 16 16 port GE RJ45

The switch is confiured as a core switch and has numerous VLANs with Layer 3 interfaces.

The switch routes across these VLANS.

When I first got here at this comapny, I noticed numerous ports in several of the blades not being used.

Today, I found out why.

When a workstation is plugged into one of these ports, the interface shows Up and I can see the input and output counters incrementing on the "sh int" for that port.

The known good workstation will not get a DHCP address or pass traffic even with a static IP config for the respective VLAN. The workstation NIC shows traffic leaving the interface, but none entering.

No logs in the switch pointing to a problem.

After verifying the physical connectivity from the four workstations I was trying to link up, I had to methodically go through the blades and find one of the unused ports that would pass traffic.

No port security issues, or configuration issues, everything is identical on the switchports, all ports have identical configuration.

Out of the four 48 port blades, two of them had sporadic ports that would not link up, all others are being used. Those two were the only ones I tried.

I had five ports on one of the blades that would not work, all others were ok, all identical config in the same VLAN and using the same workstation.

Is this a common thing with the 6509 switch or a known issue?

Jon Marshall · ‎01-03-2008

Hi Richard

Assuming these ports are faulty, I wouldn't say it is a common thing, and it shouldn't be considering the cost of 6500 switches and modules but it can happen.

In our datacentre we have a a couple of WS-X6548-GE-TX modules with a few faulty ports but due to the critical nature of the other servers connected to the modules we have simply labelled them as faulty in the interface description rather than replace them.

We have lost some ports when there was a complete power failure in our hosted data centre and the switches all rebooted.

Obviously this is assuming they are faulty and not due to cabling/power issues, are any of the modules supplying PoE ?

Jon

wilson_1234_2 · ‎01-03-2008

Thanks for the reply jon, yes these are all POE ports.

It seems like alot of ports to be faulty.

I stopeed when I had the four machines up, but there were five ports in one 48 port blade, and I didn't go through all of them.

Is there any way to determine for sure that they are faulty, othere than what I did today?

I reset one of the ports to default and reconfigured it, and still it would not link up.

Jon Marshall · ‎01-03-2008

Richard

Are the ports that are working actually supplying power to end devices ?. If so it would be worth checking that you have enough power in your chassis to power all ports.

You can run online diagnostic tests - attached is a link, although i have not done this on live switches myself.

http://www.cisco.com/en/US/docs/switches/lan/catalyst6500/ios/12.2SXF/native/configuration/guide/diagtest.html#wp1012631

Jon

wilson_1234_2 · ‎01-03-2008

That is a good thought jon, thanks.

There are phones and workstations on the majority of the ports, but not needed on the workstations I tried to uplink.

What happened was I worked through the unused ports on the blade and eventually found ports that allowed connectivity.

So I still had the workstations on the same blade that had ports that would not pass traffic.

I am wondering if a reload is in order for that switch, but I cannot because it too has critical devices on it.

Jon Marshall · ‎01-03-2008

Richard

To be honest if the ports are faulty you won't gain anything by reloading the switch anyway.

If you do get a chance though make sure you are at the console as the 6500 checks each port on each module as it boots up and displays the results on the console.

Jon

wilson_1234_2 · ‎01-03-2008

Thanks for your input jon, I appreciate it.

jcoke · ‎01-03-2008

If your SLAs permit, you might try backing the affected blade(s) out of the chassis to clear them. Other than that, time to RMA the blade(s).

glen.grant · ‎01-04-2008

You can try reloading the slot that you are having trouble with and see if that helps .

wilson_1234_2 · ‎01-07-2008

I appreciate your replys.

I have not reloaded anything, but I made an interesting discovery.

The switchport that is not working seems to be different as far as the trunking status goes.

The status of the good port is "non-trunking" and the bad port is "other", the bad port is showing as allowing no VLANs on it.

Shown are the "bad" port 33 and a good port 31:

Port Mode Encapsulation Status Native vlan

Fa3/31 off negotiate not-trunking 1

Port Vlans allowed on trunk

Fa3/31 5,250

Port Vlans allowed and active in management domain

Fa3/31 5,250

Port Vlans in spanning tree forwarding state and not pruned

Fa3/31 5,250

Port Mode Encapsulation Status Native vlan

Fa3/33 off negotiate other 1

Port Vlans allowed on trunk

Fa3/33 none

Port Vlans allowed and active in management domain

Fa3/33 none

I have not yet opened a case with TAC, but was wondering if anyone had any thoughts on this?

How do I get the switchport to be "non-trunking"?

"switchport mode access" does not work.

wilson_1234_2 · ‎02-21-2008

Reseating the blade corrected this problem.