How to check loop on the network

Anand Narayana · ‎06-19-2006

Hi,

last night i have experiance network probs, which i suspect it could be a network loop,we have 6509 as core 3750 as standby both running hsrp, 2950 is the switch were the users are connected, the connectivity lookz

6509-->3750

6509-->2950

3750-->2950

this setting is running for the past 4 months with no issue. but suddenly sterday night file transfer has become very slow, so i jus disabled the uplink in 2950 which is connecting to 3750, file transfer has become normal. i didn't make any modification, i didn't get any log in the switch related to ip address conflict of the vlanz ip address.what might have cause the problems & how to check whether looping is there or not?

i noticed in recent days in the 6509 logging buffer that the uplink to 2950 trunk down & then up, that to once in a day, switch never restarted or cable is not plugged out, but i get this message. also i noticed that port channel is on in 6509 which is connecting to 2950, actually only one cable is connected to 2950, so will that also be the probs for getting the port down & up once in a day & also for the network slow, i really don't think so because of that, but also not sure

mahmoodmkl · ‎06-19-2006

Hi

First u have done a good thing by disconnecting the redundant uplink.

Now to verify how this loop is created have a look at ur STP topology.

show spanning tree will give much info on this.check for the last TCN recieved time.if it is a most recent time then u r propbably having a loop.and check for the uplink cables it might cause due to UDLD also.

Thanks

Mahmood

leighharrison · ‎06-20-2006

Hi there,

If you have switches conencted in a loop like you have then spanning tree will block one of the ports - that is it's job and it seems as though it's been doing it just fine.

If you notice that the port between the 6509 and the 2950 has been flapping, then this is either due to a layer 1 error (cable?) or a layer 2 error (switch recieving too many bad CRC's on the interface and resetting it). This would cause spanning tree to try and reconverge - which would slow things down.

Check interface resets between the 6509 and the 2950 - is the speed/duplex ok? Also check the cable.

If you want to know if it is a spanning tree causing trouble, then you can connect a protocol analyser (www.ethereal.com) to the network and you should see much more than 1 bpdu every 2 seconds!

It also might be worth tweaking your spanning tree and putting all user/server ports into port-fast mode.

Regards,

LH

Please rate all posts