STP instance changes

Unanswered Question
Nov 5th, 2007
User Badges:

Hi folks,


In my network, RSTP is changing instance every 1 or 2 seconds. However the network is working fine. TO go in detail, we have a huge network having nearlly 242 L2 switches. All the routing is done only at core level(which is RSTP root). Access and distribution switches are acting as L2 only.

Any idea why STP is not getting stable?


Thanks in Advance.

Jack

  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 3 (1 ratings)
Loading.
Francois Tallet Mon, 11/05/2007 - 10:44
User Badges:
  • Gold, 750 points or more

Hi Jack,

Could you give more details about the instability you are seeing. The number of switches should not be really an issue. It's rather the number of instance port a given switch as to deal with that could be a problem. Your root bridge might be handling a lot of vlans on a lot of ports and might be suffering. Other possibility would be some diameter issue (if your 242 bridges were in a single ring for instance, that would be an issue;-) At last, maybe you are just seeing a lot of activity because of a single link flapping. I can't really say at that stage.

Regards,

Francois

jackthumar Mon, 11/05/2007 - 23:21
User Badges:

Hi Francois,


Thanks for reply.

we have a tree topology. from sub-access layer, data will flow to access and from access to distribution and from dist to core. all these uplinks are dot1q trunks. we have five floor, each with one dist-switch. hence ten( 2 trunk for 1 dist for redundency) links are coming to core. we have a total of 49 vlans spreading across entire data center. all these vlans are populated in core.

when ever i give show STP, its topology change count will be increased. and time since last change will be 1 or 2 sec.

so, any idea about whats causing these issues and does it have any impact on network in long run(as it is working fine at this stage).

regards,

jack



Francois Tallet Tue, 11/06/2007 - 07:27
User Badges:
  • Gold, 750 points or more

Hi Jack,

Hopefully, this does not look that bad. You are experiencing constant topology changes. Those are most likely caused by some port with no portfast configuration at the edge of the network. Each time a user powers up a pc for instance, that will generate a topology change. Considering the side of your network, it is likely this is a very frequent event. In the worst case, it would be a link flapping between switches in the network.

You can do a "show spanning-tree detail" and see the source of the last topology change a switch detected. Hoping from switch to switch using this command, you will find the exact source of at least one topology change.

Repeat as necessary;-)

Regards,

Francois

Actions

This Discussion