Weird Switch Problem...

[jorgedacosta]

Hi.

During one of our Ccna sessions with switches we noticed switches propagate vlan dats etc.

After spending some time with switches i notice sometimes empty tables are propagated.

Setting vtp client doesnt help.

What does ?

Besides this i have a secondary problem which is kinda more serious.

Im setting up a cascade of switches. ( catalyst 3500 3550).

Theres are currently 4 already installed at this customer/company .

Each switch controls a buildings pc-s.

They acquired a new building and i was called in to install the new switch,

Brand new Catalyst 3550....

Switch 1 links to switch 2 links to switch 3 etc.

Checking the running config and the vlan info i dont notice anything out of the ordinary... just one vlan on each switch ...

Plugged myself into the new switch , set a pass , made sure all there is only one vlan with everything in it .... made sure it was up ...

Connected it to the last switch....

Switch 5 blinks green , then red .... then switch 4 blinks red .. then switch 3 ... then 2 and then the whole network dies....

checking the config ... nothing has changed .... all of them remain the same ...

So why did it "crash" ?

To make this worse ... i connected my laptop to the new switch and ..

low and behold they all started working ....

.Switching off my laptop made all the switches crash .... switching it back on made them come back ...

Now theoretically , the new switch will have many pc-s attached to it

... so there isnt a serious problem of the network dieing .... or is there? what if there is a power shortage in one building ? ALL the

switches will crash.

This is not how I expected a decent switch to work.

What am I doing wrong ?

I humbly request to bask in the radiant light of your all knowing wisdom . :)

Thanks in advance .

Jorge Da Costa. (CCNA scrub)

[amitsin]

Did you the check what codes are running on the switches. Make sure that you are using the recommended codes for the switches as you are making a mix stack of switches. If you have one kind of switches then please make sure that you have the same code running on all the switches.

Please go through the following link :

http://www.cisco.com/warp/public/473/150.html

Please try to get the error what it genarates in the logs.

[robertphillips2]

Regarding your first question, usually by default Catalyst switches are set to VTP server. A VTP server can get changes to it's VLAN.dat from other VTP servers in the domain with higher revision numbers. Also, as far as I know every switch should have a VLAN 1 for management, I haven't looked at a VLAN.dat in a while to remember if it includes non-configured management VLAN.

So in summary, your first issue is one in which you just need to be careful with VTP settings, if in doubt, set a switch to VTP mode transparent before you put it on the network, that way it will not participate in VTP at all. It will not propagate bad VLAN.dats and it will not take in good ones from known VTP servers.

Good Luck

-Robert

[smoyer]

We recently experienced a similar issue. We installed a new switch for the purposes of upgraded from CatOS to IOS. The CatOS didn't recognize the GigE card installed so I needed to connect it to a known good switch via a fiber connection. I then needed the two switches to talk so I configured the VLAN on the new switch to be the same as the existing switch. When I did this, the entire network crashed. We found that VTP was set to server and when I made a VLAN change to the default VLAN, it shut down all other VLANs. We needed to set all the individual VLANs to active and set the VTP mode to transparent. All is good now. Be careful with VTP.