VPC and Spanning tree

[ANAKIN_TN]

Hi all,

Having Two nexus switchs configured on the same vpc domain and with classic STP, the number of vlan is too high causing that one of those Two switches becaume slow and reaches high number of spanning tree session, because a lot of trunk ports are configured.

So as a solution a found two options:

1- Desable spanning tree on some vlans => and i dont know what is the impact/behavior on VPC Domain switches

2-Migrate from classic spanning tree to MST spanning tree => i dont know the impact of this on VPC domain switch and if its compatible with VPC

So cisco brothers can you help ?  

[Flavio Miranda]

Hi

Do you see high CPU and Memory on the switch?  When you say "slow" how does you opserve it? ping? data transfer? Remote access? applications?

 If you could share your topology would make more sense to give an opinion. The idea of STP is loop avoidance, as you probably know, but sometimes there´s no possibility of loop, so, STP is not that important.

 MIST is a good alternative however when haldling too many vlans but, if your performance problem is really due VLANS and Trunk, then, I am not sure MIST will help you.

[ANAKIN_TN]

Hi, 

its very high CPU, some times some packet losses.

a topolgy is quite simple, the switches are in the same VPC domain and have a lot of ESXI connected with VPC link TRUNK dipatched on the Both switches .

So Desabling the spanning tree on VPC domain is not a problem ?!!!!!

[MHM Cisco World]

Never disable STP, it save your network from loop.

[Deepak Kumar]

hello,

Based on your statement, I recommend going with option 2 with MST. MST will work fine with vPC. Just think you need to consider some technical details that are similar to RSTP and add-ons as boundary ports if any.

[MHM Cisco World]

How much vlan we talking about?

are all is vpc vlan ? If not how many non vpc vlan ?

[ANAKIN_TN]

Hello ,

We are talkin about more than 1000 vlans and 118 interface trunk all vlans .

Regards 

[MHM Cisco World]

NSK# show system internal processes cpu  5 | no-more 

check the output from this command if the STP is in five top process then I with @Deepak Kumar  you need to config two MST and if not then there are other process utilize NSK CPU.

[ANAKIN_TN]

Hello , 

I dont see it , the carmelusd and lacp process are taking some good CPU whats is the name of the process STP , is STP ? 

[MHM Cisco World]

Share output here 

[ANAKIN_TN]

Hello , 

hers is the output 

[MHM Cisco World]

https://quickview.cloudapps.cisco.com/quickview/bug/CSCuo33078

check this bug 

[ANAKIN_TN]

HI MHM, 

It's excactly the problem that we have .

Should i upgrade to the last version ? 

Regards 

[MHM Cisco World]

wait, friend I check your CPU utilize.

[MHM Cisco World]

https://www.cisco.com/c/en/us/support/docs/switches/nexus-5000-series-switches/116200-qanda-nexus5000-00.html

so there are two thing here
1-bug I send previous
Workaround:Administratively shut down the interfaces which are not connected to any end devices.
2-FWM and L2rib is high because misconfig the Server side

"""MAC addresses can move between interfaces if the server Network Interface Cards (NICs) are configured for teaming/bonding, but the connected switch interfaces are not. This can be avoided if you use Link Aggregation Control Protocol (LACP) in order to bond the interfaces on both ends, or configure the server interfaces to use the NICs in an "active/standby" mode."""