File copy: svr2008->svr2008 causes 3550 to crash; all IOS

[neharris]

I've had a 3550 running on 12.1 with a single windows server 2008 machine (and a bunch of XP, 2003 and linux as well) for a while with no problems. Today I added a second windows 2008 server and when I try and copy large files across the 3550 between the 2008 servers it crashes the 3550 with a "Software Forced Crash Exception (0x0700)!" (sho tech attached) I upgraded the 3550 IOS to 12.2.44SE3 and get same results. I can repeat this at will.

I haven't sniffed a capture yet but it happens immediately in the first few kilobits (or less). There is clearly something going on between the 2 windows 2008 machines causing it that doesn't take place among the other pairs. I can copy the same files between any other pair of OSs without the crash.

(a 1.7G ISO is the file I first noticed this with)

I swapped NICs in the new 2008 machine thinking it was flow control or something but even with the second NIC its the same.

I tried different ports with no custom config with same results.

The first server 2008 is on 0/6; an IBM365 the new 2008 is on 0/1; a generic ASUS P4D. The NICs tried were the onboard SIS100M and an Intel PCIE Dual 1000MT.

Any thoughts? Does Windows 2008 send some sort of propritary discovery packet only when speaking to another 2008 server that bombs the Cisco? What's up?

There must be something wrong with the IOS code right?

This one is wild and puzzling me and I'll try almost anything requested to solve it.

thanks in advance,

->N

[Giuseppe Larosa]

Hello Nelson,

I would configure the ports for static access to disable any form on negotiation on the link

int g0/1

switchport

switchport mode access

switchport access vlan X

switchport nonegotiate

int g0/6

switchport

switchport mode access

switchport access vlan X

switchport nonegotiate

if the servers use multiple vlans configure for trunk

int g0/k

switchport

switchport trunk enc dot1q

switchport mode trunk

switchport nonegotiate

switchport trunk allowed vlan X,Y,Z

This should avoid any negotiation with the two 2008 servers.

I would try with these changes

Edit:

I used the output interpreter tool and here the answer for the sh stack portion

Note: Understanding the 'Fixed In Version' field - Sometimes the fixed-in version

specified may not be available for download from the IOS Upgrade Planner. This

is because, bug fixes are incorporated into software versions which have not been

regression tested. These builds are called interim images and will have a "."

in the build number found between the parenthesis, for example 12.3(8.1). The bug

fix will be in the next available image, and all later images. So if a bug is fixed

in 12.3(8.1), the bug fix is present in 12.3(9), 12.3(10), and so on.

REFERENCE: For more information on how IOS images are named, see Cisco IOS White

Paper.

The failure was caused by a software defect.

The stack trace decoded symbols are:

cmdr_cluster_hsrp_process_ip_addr_rsp

sc_sched_hook

process_sleep_for

bitlist_setall

switch_frame_age

jelk_frame_age_tick

switch_frame_age

ppc_process_dispatch

task_execute

Possible bug matches are listed below. Bugs with a score of .90 or more

are the most likely candidates:

Score Bugid Status Fixed In Duplicate Title

0.85 CSCsg67965 V 12.2(35)SE 12.2(37)SE None Switch may crash with https authen and radius-server retry method reor

0.85 CSCsh50635 V 12.1(22)EA10 12.2(37)SE None Swith reset when it runs out of empidb_list free entries.

You can access the bugs descriptions with a CCO account associated to service contracts

Hope to help

Giuseppe