Cisco bods - STP issue

[Mr.Stu]

Cisco 4506 x 2
Rapid PVST
12.2(18)

Problem: Spanning tree re-converges (fully 60 seconds) and causes switch-wide outage whenever a client turns off/on.

So I assume bpdu were being flooded when forwarding port went to down.

Tried portfast, bpduguard, bpdufilter. Still the same. The instant a host goes offline, spanning tree re-converges for that single vlan. I can unplug other hosts from other vlans fine.

This only happens on the switch that is NOT the rootbridge. If I swap the roots it happens on the other one.

I understand why the outage during STP re-converge, but a host cycle should not cause STP re-convergence, definably not on a portfast port.

Next steps are software upgrade (nightmare, client has no hardware maintenance), but it was working fine 3 weeks ago. No major changes noted.

This is doing my head in

Any suggestions?

 

[Mr.Stu]

tldr; ports with portfast enabled seem to be sending out TCN when they shouldnt be

 

[Styles]

Are you saying the problem is with just a single vlan?

 

[Mr.Stu]

Indeed, other portfast ports in different vlans are fine.

Root bridge is same for all vlans.

 

[Styles]

how many devices are in this vlan? have you tried creating a new vlan and putting these devices in here and going from there?

 

[Mr.Stu]

150+

Not really feasible, production network.