Showing results for 
Search instead for 
Did you mean: 
Join Customer Connection to register!

3750 management connectivity bouncing periodically

We have a weird problem with one of our remote locations. We have four switches in a hub-spoke configuration:

                          2811 Router



Switch 1 ----------  3750X Stack  --------------- Switch 3




                           Switch 2

Switch 1 is a 3560-48PS

Switch 2 is a 3560X-48P

Switch 3 is a 3560X-24P

3750X stack is a stack of two (2) C3750X-24P-S

Each of the links between the stack and the three 3560 switches are dual trunks. That is, two separate 1Gb trunks using spanning tree. They are NOT set up as etherchannels. Each switch has a management SVI (call it Vlan100) with an IP address. There is a subinterface on the 2811 router (Gi0/0.100) and the vlan is tagged through the 3750X stack.

Just last week, we upgraded the 3560 and both 3560X switches to 12.2(55)SE6 and we upgraded the 3750X stack to 15.0(2)SE. Since then, we would occasionally lose IP connectivity to the Vlan100 address for about 10-15 minutes. It's not long, but certainly long enough for our monitoring to catch it and alert us. Connectivity (pings, ssh, snmp, etc.) would then return without any corrective action done by us.

It appears as if no production traffic is being affected. That is, we continue to maintain IP connectivity to devices physically connected to all 3560 switches while the Vlan100 address is not responding. It just appears as if Vlan100 is down on the 3560 switch, but we have no way to tell that, of course.

Is there any additional MAC address table or spanning tree logging that can be done on these switches to see what happens? We've already added mac-move notification, but that hasn't logged anything when the switches appear down. STP is also set up appropriately, where one trunk interface is blocking while the other is forwarding during normal operation.

I have a TAC case open, but thought I'd throw this out for additional thoughts/ideas.