Solved: Loop Guard in Switches

hclisschennai · ‎01-23-2009

Hi Everybody,

I am analysing Cisco Switches Loop Guard Feature. It states that it prevents any Blocked Port to become Forwarding state when the port is not receiving any BPDU.

It seems this violates the STP rule. If Blocked state is not coming up when the BPDU is absent, then how STP converge and forward the traffic.

Shall anybody explain this.

regards,

R.B.Kumar

Giuseppe Larosa · ‎01-23-2009

Hello R.B. Kumar,

the context is that of a point to point trunk link.

Here with PVST the port has to receive BPDUs for all the Vlans for which it has lost the DP election on segment

The feature is good because:

you can face an event where only on Vlan X you are missing the BPDUs from other side

instead of moving the port in forwarding state for Vlan X it is better to place it in inconsistent state.

Actually the feature works at the Vlan level and has the ability to put in inconsistent state (not blocked) only one or some specific Vlans

Also if BPDUs are received again the port is automatically placed again in STP blocking state without config change.

And your understanding about unidirectional link detection is correct and STP loop guard can be a better way to deal with this issue then UDLD when using Rapid STP because it is too fast for UDLD timers.

Avoiding to form a loop on a specific vlan is important because it is enough to cause a network halt.

when all Vlans don't receive BPDUs or the link goes down it is clearly a different event.

We use STP loop guard also in combination with Rapid STP and with broadcast storm-control and we have found it to work well.

Hope to help

Giuseppe

View solution in original post

Giuseppe Larosa · ‎01-23-2009

Hello R.B. Kumar,

the idea with STP loop guard is that a non-designate port should stay in its state.

When the port stops to receive BPDUs instead of moving the port to listening through forwarding state the port is placed in STP inconsistent state.

The objective here is to avoid to have a switch reacts to missing BPDUs on the port in the standard way because this can cause the forming of loops:

if the reason of the missing BPDUs is for example a full used link the other switch still thinks to be the DP on the segment and if both put their ports in FORWARD state a loop can form and if traffic is high is not sure they are able to revert back.

see

http://www.cisco.com/en/US/tech/tk389/tk621/technologies_tech_note09186a0080094640.shtml#loop_guard

We could say that with this commmand we are influencing the possible topologies that STP can build so in this sense your note is correct.

Hope to help

Giuseppe

hclisschennai · ‎01-23-2009

Hi Giuseppe,

Thanks for your response. In fact i have surfed this link much before this posting.

I have some idea on this that it works for

1. Unidirection link failure

2. Point to Point Link

But if the Port abruptly goes to block state, and by chance if the port is not receiving the BPDUs due to some valid resons like switch power off or link disconnections, then i think it will wrongly put the port in loop inconsistence state

What you think about it?

R.B.Kumar

Giuseppe Larosa · ‎01-23-2009