Random port down for 5 sec on Catalyst 2950

[kverbeke]

Hello,

I have a strange problem on ALL of my 5 CISCO 2950 Catalyst Switches.

Totally random, a port goes down for 2-5 seconds and comes up again. People loose their connection.

System Message is:

Line Protocol on Interface FastEthernet0/x, changed state to down

Line Protocol on Interface FastEthernet0/x, changed state to up

No autonegotiantion is enabled. On swicth AND client computer, speed and link is selected manually.

It happened after upgrading IOS version. I upgraded again after the problem happened for the first time, but problem still remains.

IOS version: 12.1.(22)EA2 - CRYPTO

Can somebody please give me some troubleshooting advice?

Kind Regards

Koen Verbeke

Honda Access Europe nv

Belgium

[robphill]

What sort of devices are attached to the ports this issue is affecting? Perhaps it is something with the host device attached to the switchport that is causing the interface to go up and down. Also, run a "show post" on the switch and see if the switch is failing any of it's POST tests. Are there any additional messages logged to the syslog that might help, possibly additional errors or other conditions. Perhaps post a config of one of them without IP addresses to see if there are any configuration issues.

[kverbeke]

Hi,

All devices connected are normal Windows 2000/XP clients.

I also checked 'show post' and the result says 'PASSED', so I think that's ok.

I attached the result of the 'SHOW CONFIG' command. Maybee it helps.

Thank you very much in advance

Koen Verbeke

[Kevin Dorrell]

But are the NIC cards in the clients configured as auto or as fixed speed and duplex? If the switch is fixed, the clients must be too.

Kevin Dorrell

Luxembourg

[Kevin Dorrell]

Could you please also post a show int F0/x and show int switchport, and show run F0/x for a port that has shown this symptom?

I wonder if it's an errdisable.

Autonegotiation would not cause such behaviour. Autonegotiation takes place when the port comes on line, happens once only, and stays fixed until the port next goes offline. It does not happen on-the-fly.

Ahah! Speed and duplex are selected manually! Are they configured manually on the client computers as well? It must be auto-auto or fixed-fixed, never auto at one end and fixed at the other.

Kevin Dorrell

Luxembourg

[kverbeke]

Hello,

I rechecked the Speed and Duplex and it is never auto-manual. It's always the right combination auto-auto or manual-manual. (I already knew that rule).

It also happens on my GigaStack GBICs...

I attached some output to this message.

Thanks for helping !!

Koen Verbeke

[Kevin Dorrell]

If it happens on the Gigabit GBICs as well, that points to something rather more fundamental. That, and the fact that all 5 switches are affected. How is your power supply? Is is possible your supply has brown-outs?

I was surprised the show int did not show anything to suggest the problem, except perhaps the two interface resets.

I'll post back again if I think of anything else.

Kevin Dorrell

Luxembourg

[kverbeke]

Hi,

I also added some logging entries.

Thanks

Koen

[stalljh]

Looks like you have a loop. So possibly a spanning tree issue. If you are sure that you don't have a loop, then it must be the code. Try going back to the code that you had before you upgraded and see if the problem is gone.

[mhussein]

Hi,

I noticed this message:

%LINEPROTO-CLUSTER_MEMBER_1-5-UPDOWN: Line protocol on Interface FastEthernet0/23, changed state to up

and this config line:

interface FastEthernet0/23

spanning-tree portfast

If they are both on the same switch, then the loop is created by connecting a switch to a "portfast" interface. You need to remove "spanning-tree portfast" config from all ports that connect the switches.

Regards,

Mustafa

[kverbeke]

Hello,

F0/23 is normal port to which a client computer is connected. PortFast is enabled on all 'client'- ports.

My GigaStack GBICs are not PortFast...

Now, I changed something on the physical configuration of my GBICs (I swicthed them and restarted my switches)...

Maybee this helps.

I keep you informed

Thank in advance !

Koen

[amikat]

Hi,

There are known issues with gigastack gbics flapping.

Will you please check that all your gigabit gigastack interfaces are set to negotiation auto and duplex auto (which is default) and if possible remove and reinsert the gigastack gbics and see if there is any improvement.

Best regards,

Antonin

[kverbeke]

Hello,

I just changed something on the physical configuration of my GigaStack GBICs... (I switched them...) I'll keep you informed about pergress...

Thank you very much for the tip,

Koen