Switching problem 4006 SupervisorII CatOS 8.4.4GLX

[audejacq]

Hi,

We just upgrade the CatOs of our 4006 and discover that we were in the field notice problem because of the Lemans error. The problem is that now were having some workstation that can ping 192.168.67.23 but not .24 ??? If we disconnect the network cable for at least an hour sometimes it's ok sometimes not. The next day the worstation is ok and it's the PC in the next office that goes of but he can ping .24 but not .23. Can the Lemans error cause this??

Help me plz it's a priority #1 here

[Prashanth Krishnappa]

Lehmans is a stub ASIC on oversubscribed cards. I would not blame it yet.

What is 192.168.67.23 and .24?

Which switch do they physically reside on?

Which VLAN are they in?

Which VLAN are your workstations in?

Did you check for any port errors?

In the broken state some troubleshooting needs to be done.

[audejacq]

Hi,

The 2 address are servers. The switch where are the workstation connected are 3500XL, 3550, 2950. All the workstation are in the same VLAN 200 and the management is int the VLAN 70. We've done further check and it seems that the arp table do not flush correctly.

Thanks

[glen.grant]

When this happens is there anything that shows up in the logging buffer ? When you reloaded did all the cards come online right ? Do a show test 0 and show test switch-process memory detail and see if any errors show up .

[audejacq]

Hi Glen,

There's no error in the 2 commands. But the logging buffer show the Lemans error.

%SYS-4-P2_WARN: 1/Lemans 6-5(6/17-20) - consecutive management requests timed out.

[Prashanth Krishnappa]

Are any of the machines you ping from/to on ports 6/17-20? Do you see these errors during the time when you are having connectivity issues? If so, it could be the cause. Check the following page for troubleshooting this issue

http://www.cisco.com/warp/public/473/astro_45640.html#topic2

[audejacq]

There's no machines in the ports 6/17-20. The problem of connectivity is always there but appear and disappear randomly. And the Lemans errors coming every 10 minutes and still.

[Prashanth Krishnappa]

Check to see if you are seeing any huge broadcasts/Unicast flood in your network. Plug in a machine with a sniffer into a port without any SPAN and see if you see traffic which the port is not supposed to see.

[audejacq]

For the workstation in problem, it seems that when they clear there arp table, they resend arp request, I can see it from another workstation using ethereal. But with other tools like ethereal I've done the same thing on the server and no arp request from this particular PC arrived to the server. It seems that we have a mini flood of arp request coming from everywhere.

[glen.grant]

Is it always the same card asic 6-5 that has this? If so then I would lean towards the line card being faulty , if it is changing then I would suspect the supervisor .

[audejacq]

No, it's happening usually on server witch are connected in blade 2, 3, 4, 5, 6 on GBIC LX/SX and baseT. The workstation are connected on switch that are connected in the 4006.

[audejacq]

We finally rollback through CatOS 8.1.1 last night, we still have the Lemans error but it seems that all the other problems are gone. Thanks you for you're help.