Cisco 4506 crash and recover automatically

[wayne.leung]

Lately one of the Cisco Catalyst 4506 switches here at my site has been experiencing an interesting phenomenon in which the switch will crash consistently at a certain time during the day and recover automatically by itself. During the crash all the users connecting to the switch will be knocked off but will be reconnected automatically.

Having eliminated the possibility of foul play and power failure I have spent some time going through all possible diagnositc commands such as "logging buffer" ,"sh test 1", "sh tech-support", "sh log", "sh nvram 1" and what's most interesting is that I haven't seen any power resets during the time when the users reported that the switch crashed (that they got knocked off the network).

Frankly, I am out of ideas and I am hoping someone with a similar experience can shed some light on this topic. For your information, here is a quick breakdown of the modules in the chassis:

Cisco CatOS version:7.4 (3)

Cisco Cataly 4506 Chassis

Super Engine II (WS-X4013)

48 Fast Ethernet switch port blade (WS-X4148-RJ)

Could this be a CatOS issue ?

Thanks.

[Prashanth Krishnappa]

Paste the output of

sh crashdump 1

sh nvram 1

dir bootflash:

Also after you capture the above, clear the crashdump using "clear crashdump 1"

[wayne.leung]

Here is the output of "sh nvram 1"

PS1="rommon ! >"

?="0"

DiagBootMode="post"

vitConsecTimeouts="1"

vitTimeouts="14"

MemorySize="64"

ResetCause="20"

AutobootStatus="success"

Here is the output of "dir bootflash:"

-#- -length- -----date/time------ name

1 4473312 May 22 2003 02:12:41 cat4000-k8.7-4-3.bin

11124128 bytes available (4473440 bytes used)

[wayne.leung]

Please see the attached for the sh crashdump 1 because of the size.

[ddrodge]

Wayne,

Any resolution to this? We have a C4506 with a SUP IV running IOS 12.2(20) with the same problem. System reloads every morning around 7:00 AM

Dereck

[wayne.leung]

Hi Dereck,

Last time my problem with the switch was actually tied in with the failure of the UPS battery. If you see a "ResetCause=”20”" in your "sh nvram 1" then you should examine your UPS or power source for your switch. If not, I would suggest do a knowledgebase search on your resetcause number on Cisco's site.

Hope it helps.

[Kevin Dorrell]

In the case of the 4006, it needs two power supplies to run. I use all three slots, so any one PSU can fail. But I have seen that if you go down to one PSU only, it still runs the supervisor, but can kick out the line cards to save power. But I don't know how the 4506 behaves in this regard.

Kevin Dorrell

Luxembourg

[Prashanth Krishnappa]

Can you paste the output of sh ver, sh mod and a sh platform crashdump?

[ddrodge]

Thanks for the reply. Further analysis showed that the UPS battery was dead and the building was (and is) getting power spikes. The UPS is being replaced with another and the problem has gone away. It is interesting that the sh platform crashdump shows:

Last powerfail: 09/15/2004 09:10:24

Current time: 09/15/2004 09:21:16

vet sh versions shows:

SW9 uptime is 5 minutes

System returned to ROM by reload

and not System was returned by power on.

Dereck