2960S to SG200-08P - Network Outage

[jsibert69]

First a little run down.  3 2960s DMF (FlexStack) fiber to IDF1 (Stack of 2 2960s) and fiber to IDF2 (Stack of 3 2960s).

I have 2 SG200-08P switches as slaves off of IDF2 in my building.  I installed them without configuring anything on the SG200's.  They ran okay for a couple months.  Yesterday I had a complete network outage at IDF2.  No traffic was getting to or from the MDF.  After trial and error, I determined that it was the port connected to the SG200 that was bringing everything down.  I disconnected it, put in a dumb SG100 in it's place and the network came up.  The next morning I come in, network outage again.  Again I took an SG100 and replaced the second SG200, network came back up.

I took the 2 units and logged into them from my PC, reset the units, disabled STP, and reinstalled the units.  The network is up and running.  I have experienced this problem when I had a generic Smart Intellinet 24 port switch connected some time back.  It would take down the segment it was connected to.

Can someone explain why this happened and if disabling STP on the SG200's was indeed the solution to my problem or was it just the resetting of the units that fixed the issue.

This was quite frustrating. 

My setup on the 2960 to the SG200 is:

IDF2-Sw1-Rk1#show ru int gi 1/0/7

Building configuration...

Current configuration : 202 bytes

!

interface GigabitEthernet1/0/7

switchport trunk allowed vlan 1

switchport mode trunk

macro description cisco-to-ems-in-aquarium

spanning-tree portfast

spanning-tree link-type point-to-point

end

IDF2-Sw1-Rk1#

Thanks for the help and input.

Jeremy

[Tom Watts]

Hi Jeremy, the Sg100 has no concept of spanning tree so it shouldn't be the solution. But the one thing to note, you do have port-fast enabled on the 2960... which shouldn't be enabled interconnecting switches. I'm not sure the 2960 support spanning-tree port-fast trunk configuration but if you're using port fast, that would be the correct setting.

As for the port configuration, it is also kind of interesting to note, gi1/0/7 of the 2960s is sending VLAN 1 tagged. Did the sg200-8 use VLAN 1 tag? The Sg100 has no concept of VLAN so the tag VLAN could have been confusing, perhaps.

In the end, I wouldn't be surprised if some sort of BPDU was received causing a port to disable. We would probably need the log messages from the 2960 and the sg200-8 to know any better.

-Tom
Please mark answered for helpful posts

[jsibert69]

No, I'm not doing anything special with them yet.  I do have a voice vlan that I have been configuring but I haven't made it that far to deploy it throughout the building yet.  The SG100's I know are just dumb, so whatever the issue is has to be something that the SG200/2960 didn't like.  It strange though, when it took everything down, local communication to IDF2 was fine, I could talk within that switch cluster but it seemed to kill all traffic to and from the MDF which is coming off GI 1/0/52 over fiber.  Does that make sense?  Would it be the MDF cluster that shut off or IDF2, that I don't know and haven't been able to determine.

[Nagaraja Thanthry]

Hello Jeremy,

SG200 is a smartswitch with Spanning Tree capabilities. From the looks of it, the issue seem to be due to the 2960 identifying a loop in the network and shutting down the port. When we look at the configuration of the 2960, you have configured "spanning-tree portfast" which essentially forces that port to go into "forwarding" mode as soon as a device is connected and not go through the entire spanning-tree calculation. Portfast is configured on ports where you have an end device or a dumb switch not capable of participating in spanning tree calculations. You might want to consider removing the "portfast" configuration on the ports where the SG200 is connected and reenablng spanning-tree on SG200 and then monitor the network.

Nagaraja