Possible 3750 stackwise issue ??

[andymh]

I've just upgraded from 2 x 3750 core switches with a 2Gb Etherchannel to a 3 x 3750 core via the stackwise cables. The new stack appears to work fine, and has the backbone running at full speed.

My problem is that since this work has been carried out a number of remote sites are having major connectivity issues with a proxy server and groupwise server within the stack. Ping times from the remote sites are low (15ms which is fine).

I am using MRTG to monitor the remote links and there are frequent blips where it will miss 30 secs of logging.

What I can't work out is whether the new stack is causing the problem or whether it is something else. I guess the best way to eliminate would be to return to the previous setup, but that's a last resort, as I'd prefer to see if anyone else has had issues with this first ??

Also is there any way to troubleshoot the stack using n/w assistant in a meaningful way

, rather than just look at port stats for each port.

(p.s. The HQ router and the Groupwise server are on the same 3750 switch)

[Sriharshaa Prabhakar]

Hi,

Can u attach the show technical support output from the switch?

Regards,

Sriharshaa Prabhakar

[andymh]

I tried with sh tec page, but the output is so vast that I can't cut and paste the whole amount as it overflows the screen buffer on hypertrm.

Is there a way to pipe it to flash and tftp it off or similar ?

[royalblues]

When you start hypertrm, go to the transfer window and enable the capture to a particular.txt file.

After this you can issue the sh tech command. Evrything will be saved in the text file which can be uploaded

Narayan

[andymh]

Cheers for that. See attached !!

[sourabhagarwal]

GigabitEthernet1/0/6 interface is taking CRC and collisions. can you clear counters and check if these errors and collision rate is increasing. If yes, there could be speed/duplex issue between this interface and interface to which it connects or could be due to a bad NIC, faulty switchport, bad cable. I can see from config that speed/duplex setting on Gi1/0/6 is set to AUTO. make sure other end is also set to AUTO. we need to fix this issue first.

I also observed that the following processes are causing excessive CPU usage:

PID CPU Time Process

202 26.51% Virtual Exec

The virtual EXEC process handles virtual type terminal (vty) lines, such

as Telnet sessions on the router. Issuing commands that generate long output (such

as 'show tech-support'), or if the debug output has been redirected to the vty

(using the terminal monitor or the no logging console command), the amount of CPU

resources used by the virtual EXEC process increases.

Make sure all debugging commands in your router are turned off by issuing the undebug all or no debug all command.

hope to help ....

[glen.grant]

The only thing I can see that I might try is to pull that span session off and see if that has any effect , you are basically spanning the whole switch would could cause added overhead . Not sure what you are doing with that span session .

[andymh]

The RSPAN session was supposed to allow me to run my LAN Observer to try to see what was going on. It doesn't work, but taking it off previously didn't help either.

The Gi1/0/6 I have disabled at the server. It has 2 x network cards. This hasn't made any difference.