Weird Problem with cisco 2950 and 3560-X

[alexandremonteiro1]

Hi,

I have several switches, three of them are managed from Cisco.

All switches are connected to the Backbone ( 3560-X )

the 3560-X is our backbone and it is connected to the GW/FW and to the 2950 and other 3560 with trunks.

If for some reason i disconnect or restart the 2950, the network on the Ciscos goes down.

2950 is the STP master ( im going to alter this today )
If i disable STP on vlan1 on the 2950 the vlan 1 on the three switches goes down.

The Switch 2950 have a high ping but has only about 5 devices connected and with low traffic, if i ping the device the ping is good.

My questions are:

why do STP protocol disrupts networks traffic when is altered?

Why the 2950 have so high ping?
whan can i do to improve or remove that disrupts on STP change?

thanks for the attention,

AM

[Elton Babcock]

If you are running basic STP and the 2950 is your "master" or better called the root bridge, then STP needs to re-calculate its links and also go through the STP election process. This can take I believe up to 45 seconds.

As for the high ping times, what VLAN is being used for the management of that switch and is your management PC on this same VLAN? If not what is doing the routing of the traffic? Most likely the GW or the 3560-X switch.

If you could give a little clearer picture of how things are setup that would be helpful.

Sent from Cisco Technical Support iPhone App

[alexandremonteiro1]

Hi Elton, thanks for the reply.

The routing of the traffic is done by the FW/GW a Watchguard XTM 515.
My Workstation is on Vlan 20 and all the switches are managed by the vlan 20.
I have several vlans:

I have a seperate network for servers connected to one interface of the Watchguard

1 and 20  - corporate network

10 - Voice vlan

30 and 35 - Unsecure and GuestWireless vlans
40 and 45 -  Cluster Failover vlans

My need for using vlan 1 is that we have several unmanaged switches, and cannot aford to buy more cisco atm.

The switch 3560-X is l2 ( cluster vlans are only configured here )

Now the ping to the 2950 ( on same vlan ) is ok, but friday was around 600ms to 2000ms with spikes of timeout's.
So you are saying that if i change the STP root/primary, ill loose conectivity on all the vlans affected for about 45secs?

If you need part of the switches configuration please say so.

sincerely,
AM

[Elton Babcock]

Most instances of STP run per VLAN and it is actually called per VLAN spanning tree protocol or PVSTP.

Usually your most backbone switch should be the root as that is your most powerful and you want the traffic traversing that switch and not an access switch in some cases.

You can change the root per VLAN or do it at once for the entire switch. In each case STP will need to go through its stages for listening and learning before it will forward traffic. (Usually about 45 seconds)

As for the high ping times that sounds like there was a physical loop somewhere in the network and your switches were getting flooded with broadcasts. I have seen that exact behavior when a loop exists.

Sent from Cisco Technical Support iPhone App

[alexandremonteiro1]

I have checked some ports and five of them on the Backbone are with high traffic precisely 31% ( transmit only ) usage on each one of them.

The interesting part is that one is connected to a server that is offline and the other is connected to a Cisco IP Phone only, the other three are connected to one other switch and to two access points.

I disconnected one of the Access points and the result is the same, only that the traffic on that port is null because it is disconnected.

On that switch there are people with wireless and with cable, could this be the source of the problems?