Spanning-tree portfast

[darren-carr]

Guys,

Across all of my access layer switches (including trunk ports) we have the configuration 'spanning-tree portfast' enabled. I fully understand and appreciate the need of having this on access layer switchports but is it really necessary to have it applied on a trunk?

Thanks

Darren

[lejoe.thomas]

No as you pointed, it should not be applied to a trunk, considering that normally other networking devices are connected to the trunk rather than an end station.

HTH

Lejoe

[darren-carr]

Hi Lejoe,

Thanks for the prompt response.

You would also not expect to see this configured on a logical Port/Etherchannel either?

The guy who implemented the network has just flashed it across all interfaces both physical and logical! and Im not just trying to tidy up. Again I don't believe they should be configured here

Thanks

Darren

[Jon Marshall]

Darren

The only place i would expect to see "portfast trunk" is on a trunk connection to a NIC on a server that can also run 802.1q.

Jon

[John Blakley]

I agree with Jon. In fact, I asked almost the same exact question the other day. :-) (Mine was more along the lines of uplinkfast and why I would use it instead of portfast trunk.) Seemed to do the same thing, but very different. Actually, it's different because of the way spanning-tree handles the cost of the link and the bridge priority, but that's another thread. :-)

--John

[lejoe.thomas]

Jon,

Thanks for pointing that out.

Lejoe

[lejoe.thomas · Accepted solution]

Darren,

Again the logical port channel/etherchannel is an aggregation of physical ports, which will be connected other networking devices or server supporting aggregation, rather than an end station, so you can safely remove it. In fact, it shouldnt even be present. Normally portfast on a port, always think end station would be connected to that port, which would not generate any BPDUs.

HTH

Lejoe