- Duplex mismatch—When one side of a link is half-duplex and the other is full-duplex. This causes late collisions and FCS errors.
- Unidirectional link failure—The link is up but data flows only in one direction. It can cause loops.
- Frame corruption—Physical errors on the line cause BPDUs to be lost, and the port incorrectly begins forwarding. This is caused by duplex mismatch, bad cable, or cable too long.
- Resource errors—STP is implemented in software, so a switch with an overloaded CPU or memory might neglect some STP duties.
- Port Fast configuration errors—Connecting a switch to two ports that have Port Fast enabled. This can cause a loop.
- STP tuning errors—Max age or forward delay set too short can cause a loop. A network diameter that is set too low causes BPDUs to be discarded and affects STP convergence.
Identifying a Bridging Loop
Suspect a loop if you see the following:
- You capture traffic on a link, and see the same frames multiple times.
- All users in a bridging domain have connectivity problems at the same time.
- There is abnormally high port utilization.
To remedy a loop quickly, shut redundant ports and then enable them one at a time. Some switches allow debugging of STP (not 3550/2950) to help in diagnosing problems.
What to Use Where
Confused by all the acronyms and STP features? Figure 3-3 shows the STP features you might use in your network and where you might use them.
No comments:
Post a Comment