SPTBPDUTX - High Utilization

[a-rohilla]

Hi

In my core switch when i do sh proc cpu i see overall utilization of the sup(SUP1A with MSFC-3) 30-34%, the 2 contributing processes to this high utilization are sptbpdutx(14%)and sptbpdurx(15%). What is the cause of high utilization of these 2 processes and how to detect it.

Regards

[a-rohilla]

Hi Vijay

When i see output of command "sh spantree statistics 1/1 2" then i see config bpdu's xmitted (port/VLAN) increasing every sec while config bpdu'd recieved and TCN bpdu's not increasing, and this happens at each trunk port of core switch, hellow timer is set to default 2 sec, what i feel is i should increase the hello timer since it is sending config bpdu's every 2 sec. what do u feel ?

Regards

[vijayasankar]

Hi,

Those two process are related to Spanning tree.

You need to check the stability of spanning tree in the network to trace the root cause.

-VJ

[glen.grant]

If you are doing a lot of trunking then "manually" prune off any vlans that are not needed across trunked links , if you have it wide open then the switch has to allocate spanning tree instances to all vlans allowed which drives up the cpu...

[a-rohilla]

But can i increase the hello timer from 2 sec since config bpdu are sent every 2 sec which is causing high config bpdu transmit thru the trunk.

Regards

[glen.grant]

I would not mess with the default values , if spanning is causing high cpu there is a reason for it .

[a-rohilla]

Pruning is the only thing which i can do, is there anything else also which i should look for to reduce this utilization.

Regards