Potential spanning tree issue

andrewrocks · ‎10-17-2011

Hi All,

I have a Cisco 3560 running as a Level3 device in my network running 10 VLANs and routing between most of them (nothing complex with ACLs) and running spanning-tree mode pvst. The main network is run on a netgear GS748TPS stack of three switches running MSTP.

I have just bought an additional 3560 and a 2960 to plug in. I have set them up with IP addresses and then plugged them into the netgear. This brought the whole network down until I unplugged the new switches.

I have confirmed the IP addresses aren't duplicated and that DHCP is not running on the switches so I can only assume it's something to do with DHCP. I cannot afford for the network to go offline again, so is there anything I should check? Am I running incompatible spanning tree methods between the netgear and cisco devices?

Thanks for any and all help.

Regards

Andrew.

kozorezdi · ‎10-18-2011

Hello Andrew,

have you checked syslog on all devices? I suppose it should be the first step...

--

Dmitry

Thotsaphon Lueangwattanaphong · ‎10-18-2011

Hi Andrew,

Are you planing to do a migration? If not, you just do a routing between those 2 networks to get rid of spanning tree problem. I'm not sure that netgear can do "routed port/Layer3 interface" to do that. However, please provide more information how you connect them togeher, including a brief diagram.

In your case it might be spanning tree problem(root election). Did you just plug 1 cable to connect them together? It's a good idea to see what the logs are.

HTH,

Toshi

andrewrocks · ‎10-18-2011

Dmitry: I'm new to Cisco equipment so am unsure how to check the syslog, can you please elaborate a bit?

Toshi: I'm not planning a migration, no. I don't have the budget for this. The plan is to keep the netgear stack and two other netgear switches for the desktops / telephone and use the Cisco switches for the routing between VLANs and for all mission critical servers / devices.

The basic layout is:

Each switch is only connected with 1 cable currently, and as I said, I have checked for duplicated IPs etc.

Thanks for your help

Andrew.

Peter Paluch · ‎10-18-2011

Andrew,

I am somewhat worried about the STP type mismatch in your network. I know too little about your network and what happened, but with respect to STP, there are a few things that can go wrong. Personally, I would strongly vouch for using MSTP on all your switches, using identical MSTP region configuration.

Can you give us more details about what you mean by "bringing the entire network down"? Did it exhibit symptoms of Layer2 loop? Did the network partition into several parts that could not communicate between themselves, only within these parts? Did any workstation complain about anything unusual? Any further information beyond "the network went down" is very helpful.

Best regards,

Peter

andrewrocks · ‎10-18-2011

Hi Peter,

I thought that MSTP was the netgear equivalent of pvst?

When I say "the network came down" I mean that I was not able to get from the netgear to the cisco. I have two PCs plugged into the netgear stack switch and when the network "went down" I was no longer able to ping the cisco 3560 nor any other device on the netgear. Because of the nature of our business I was forced to pull the plug on the new switch rather than spend time working out if I could access anything else.

How do I examine the syslog?

Regards

Andrew.

Peter Paluch · ‎10-18-2011

Hi Andrew,

I thought that MSTP was the netgear equivalent of pvst?

Oh, not at all. (R)PVST+ is a strongly different incarnation of STP when compared to MSTP. Using the same version of STP - MSTP in your case - is almost a must. I would suggest not going further before making sure that all switches run MSTP and use the same MSTP region configuration.

Would you mind posting the MSTP part of configuration from the Netgear switches? I would be happy to create a corresponding MSTP configuration for your Catalysts.

The logging messages on Cisco switches can be displayed using the show logging command. However, this assumes that the buffered logging has been enabled - sometimes it is by default, in other IOSes it is not. This buffer gets erased after switch is powered off or reloaded.

Best regards,

Peter

andrewrocks · ‎10-18-2011

Hi Peter,

That makes sense. I think I was under some false impressions of spanning tree.

Thanks for your help

Regards

Andrew.

Peter Paluch · ‎10-18-2011

Hi Andrew,

The images are perfectly readable, however, I am missing the crucial MSTP configuration elements, which are:

MSTP Region Name
MSTP Configuration Revision
MSTP Instances and VLAN mappings

Is it possible to dig out these elements somewhere in the GUI or command line (if the Netgear has any)?

Best regards,

Peter

andrewrocks · ‎10-18-2011

I'm afraid the netgear isn't advanced enough for a CLI.

I'm not sure but isn't the "32768-C0:3F:0E:39:AD:3F" the region name?

I'm sorry, I'm trying to find the information and will keep digging.

Thanks for your patience.

Andrew.

Peter Paluch · ‎10-18-2011

Hi Andrew,

That string appears to be the Bridge ID - the composition of a priority and the MAC address of the switch. And hey, you do not need to apologize - there's no reason, okay?

Perhaps some operations manual will contain information about the default MSTP configuration for the Netgear - I assume you have not changed it.

Best regards,

Peter

Alexander Maroukian · ‎10-18-2011

Hi Andrew,

Is your mstp working now? Or how does it work now? Maybe it is working without vlans defined, or it includes all by default? I do not see any configuration on the screenshots you have provided. I have looked at the manual of your netgear device. I cannot see the vlan mapping in the VLAN ID cells next to each mst id.

The revision number is 0, as I think this is the revision number of the current stp configuration.

Best regards,

Alex

EDIT: Peter do you think it is possible that "configuration name" on "stp configuration" tab which is the same as the mac address in this case to be the name of the region?

andrewrocks · ‎10-18-2011

Hi Both,

Thinking about it, I don't know if it is really working now. I'm guessing it isn't seeing as I'm having these issues.

Should I just disable the Spanning Tree on the Netgear and rely on the Cisco to keep the inter-vlan state and mission critical devices ok?

Regards

Andrew

Peter Paluch · ‎10-18-2011

Hi Andrew,

I got it. The MSTP Configuration Name and Configuration Level are available at the Basic/STP Configuration page.

My recommendation is to unify these configurations across your switches. You will have to configure these data on all your Netgear switches identically. Use the following data:

Configuration Name: MST
Configuration Revision: 1

Furthermore, it seems that your switch supports two MST instances in addition to MST Instance 0 (called CST in your configuration), and that all VLANs are currently mapped to the CST, not to the two additional instances. That is fine for your needs, and should not need to be modified.

So the only configuration changes I suggest doing is modifying the Configuration Name and Revision on all your Netgear switches (please do this in a quieter period, as modifying these settings may cause short connectivity outages). Also make sure that all your Netgear switches are configured to run MSTP.

Your Catalyst switches should then be configured as follows:

configure terminal

spanning-tree mst configuration

name MST

revision 1

instance 0 vlan 1-4094

exit

spanning-tree mst 0 priority 61440

spanning-tree mode mst

end

Do you believe these modifications would be possible?

Best regards,

Peter

andrewrocks · ‎10-18-2011

Hi Peter,

Ah I see, I will try and do this.

Can I ask, does this mean that EVERY switch must run MSTP? I ask because our warehouse has several dumb (i.e. cheap) switches hanging off access ports and these switches have no spanning tree at all.

I will try and find a quite time to do this and report back.

Thank you for your help.

Regards

Andrew.