UDLD Problem after switch reload

[gbonivento]

I have the following problem with

Situation:

- 2x 3750G-12S Distribution-Switches (DS) and several 3560/2960 Access-Switches (AS)

- redundant Fiber optic uplinks between AS and DS

- Cross-Stack Etherchannel config on all uplinks

- UDLD aggressive mode configured on all uplinks

Problem:

- when I reload the DS and the switch comes up again, I've lost the connection with several (not all) AS

- looking at the AS log I found the following error message: "%UDLD-4-UDLD_PORT_DISABLED: UDLD disabled interface Gi0/1, aggressive mode failure detected"

Cause:

- in the log of the AS I don't see intf Gi0/1 go down during the reload of the DS

- this means the DS didn't shuts down all its ports during the reload

- but this also means that UDLD on AS didn't know about DS reload, and therefore didn't resets counters

Workaroud/Solution

- I'm now configuring errdisable recovery on all AS

So I'm forced to configure errdisable recovery, even if I don't want.

Any other solutions/workaround?

For Cisco guys: what do you think about my case: it has to be considered a Cisco bug?

Regards, Gio.

[Nagaraja Thanthry]

Hello,

How are the DS and AS switches connected? Do you have any patch panel or

signal booster in between? When you reload the switches, the switch will

turn-off all the ports for a short duration. You can see that in the console

logs when the switch was rebooting. So, if in your case it is not happening,

it could be a hardware issue or a software issue. Can you remove one of the

connections from the DS and see if the corresponding AS reports link down?

Regards,

NT

[gbonivento]

Hi Nagaraja,

yesterday evening I reloaded two more Distribution Stacks (upgrade from 12.2(50)SE3 to 12.2(55)SE due to a SW Bug) and I experienced again the same problem.

Here below an example of the logging on the AS:

000329: Sep 13 14:43:08: %UDLD-4-UDLD_PORT_DISABLED: UDLD disabled interface Gi0/1, aggressive mode

failure detected

000330: Sep 13 14:43:08: %PM-4-ERR_DISABLE: udld error detected on Gi0/1, putting Gi0/1 in err-disable state

000331: Sep 13 14:43:09: %LINEPROTO-5-UPDOWN: Line protocol on Interface GigabitEthernet0/1, changed

state to down

000332: Sep 13 14:43:10: %LINK-3-UPDOWN: Interface GigabitEthernet0/1, changed state to down

000333: Sep 13 14:43:11: %UDLD-4-UDLD_PORT_DISABLED: UDLD disabled interface Gi0/2, aggressive mode

failure detected

000334: Sep 13 14:43:11: %PM-4-ERR_DISABLE: udld error detected on Gi0/2, putting Gi0/2 in err-disable state

000335: Sep 13 14:43:12: %LINEPROTO-5-UPDOWN: Line protocol on Interface GigabitEthernet0/2, changed

state to down

Note: no "interface GigabitEthernet x/y changed to state down" message before the errdisable message, even if the device on the other side of the uplink was reloaded.

I'm sure that I have no active components between the AS and the DS.

Attention: I reviewed all logs of my lasts reload activities and one interesting things comes out: each time I experienced the UDLD Errdisable problem it was on Gigabit COPPER uplinks.

So it seems the 3750 Distribution switches didn't put in down state the Gigabit Copper SFP modules ...

[Simen Ringstad]

I had the same problem as you tonight while upgrading a WS-C2960-24TT-L. When it rebooted, the neighbour-switch put the 10/100/1000BaseTX interface in err-disable state.

Here's the log from the neighbouring WS-C2960-24TT-L:

Apr 30 00:41:33.756 UTC: %UDLD-4-UDLD_PORT_DISABLED: UDLD disabled interface Gi0/2, aggressive mode failure detected

Apr 30 00:41:33.756 UTC: %PM-4-ERR_DISABLE: udld error detected on Gi0/2, putting Gi0/2 in err-disable state

Apr 30 00:41:34.763 UTC: %LINEPROTO-5-UPDOWN: Line protocol on Interface GigabitEthernet0/2, changed state to down

Apr 30 00:41:35.778 UTC: %LINK-3-UPDOWN: Interface GigabitEthernet0/2, changed state to down