Host [mac addr] flapping between port A and port B on Cat2948G switch

[patrick.zhang]

Recentlly,I find error message in my Cat2948 switch,

The topo is:CBX500(Lucent switch) connect to port 2/1 of Cat2948,another switch connect to port 2/5 of cat2948,and few hosts connect to other cat2948 ports.

Many times,I found error message which said:

%SYS-4-P2_WARN: 1/Host 00:60:d1:01:5b:f8 is f

lapping between port 2/5 and port 2/1.

I checked the physical connection,no loop,and STP enabled in cat2948.Mac address,00:60:d1:01:5b:f8,is the addr of CBX500.

When flapping happened,3 to 10 packets lost between cat2948 and CBX,this serious affectted our applications.I have refered the tech files about Cat4000 serial switch error message,useless.

sh proc cpu

CPU utilization for five seconds: 5.46%

one minute: 6.00%

five minutes: 6.00%

Cpu utilization seem to be ok.

Can you give me a solution?

Thank you.

[milan.kulik]

Hi,

this error message usually appears when STP topology changes.

Is STP running on all your switches? Have you noticed any port going down/up at the time of the error?

Try to enable SNMP traps or syslog messages for STP TCN and check if you receive STP TCN alarm next time the error appears.

Another possibility is that both your Lucent switches are somehow misconfigured to use the same MAC address 00:60:d1:01:5b:f8.

Regards,

Milan

[patrick.zhang]

Hi,

Thank you for help!

But I think,cat2948 has STP enabled,host with mac 00:60:d1:01:5b:f8 is direct connected to cat2948,so at least,it should not be learned in different port in same time no matter other devices STP enabled or not.

"Another possibility is that both your Lucent switches are somehow misconfigured to use the same MAC address 00:60:d1:01:5b:f8.",what this mean? and how it can?

This error message come out one time per 1 or 2 hours,in cat2948,no port up/down log,no any error in these two ports.

sh spantr

VLAN 1

Spanning tree mode PVST+

Spanning tree type ieee

Spanning tree enabled

Designated Root 00-02-fd-ae-c4-8d

Designated Root Priority 32768

Designated Root Cost 219

Designated Root Port 2/5

Root Max Age 20 sec Hello Time 2 sec Forward Delay 15 sec

Bridge ID MAC ADDR 00-0b-fc-e8-60-00

Bridge ID Priority 32768

Bridge Max Age 20 sec Hello Time 2 sec Forward Delay 15 sec

Port Vlan Port-State Cost Prio Portfast Channel_id

------------------------ ---- ------------- --------- ---- -------- ----------

2/1 1 forwarding 19 32 disabled 0

2/2 1 forwarding 19 32 disabled 0

2/3 1 forwarding 19 32 disabled 0

2/4 1 not-connected 100 32 disabled 0

2/5 1 forwarding 100 32 disabled 0

2/6 1 not-connected 100 32 disabled 0

2/7 1 not-connected 100 32 disabled 0

2/8 1 not-connected 100 32 disabled 0

2/9 1 not-connected 100 32 disabled 0

2/10 1 forwarding 19 32 disabled 0

2/11 1 forwarding 19 32 disabled 0

2/12 1 forwarding 19 32 disabled 0

2/13 1 forwarding 19 32 disabled 0

2/14 1 not-connected 100 32 disabled 0

2/15 1 forwarding 19 32 disabled 0

2/16 1 forwarding 19 32 disabled 0

2/17 1 forwarding 19 32 disabled 0

2/18 1 not-connected 100 32 disabled 0

2/19 1 not-connected 100 32 disabled 0

2/20 1 not-connected 100 32 disabled 0

2/21 1 not-connected 100 32 disabled 0

2/22 1 not-connected 100 32 disabled 0

2/23 1 not-connected 100 32 disabled 0

2/24 1 not-connected 100 32 disabled 0

2/25 1 not-connected 100 32 disabled 0

2/26 1 not-connected 100 32 disabled 0

2/27 1 not-connected 100 32 disabled 0

2/28 1 not-connected 100 32 disabled 0

2/29 1 not-connected 100 32 disabled 0

2/30 1 not-connected 100 32 disabled 0

2/31 1 not-connected 100 32 disabled 0

2/32 1 not-connected 100 32 disabled 0

2/33 1 not-connected 100 32 disabled 0

2/34 1 not-connected 100 32 disabled 0

2/35 1 not-connected 100 32 disabled 0

2/36 1 not-connected 100 32 disabled 0

2/37 1 not-connected 100 32 disabled 0

2/38 1 not-connected 100 32 disabled 0

2/39 1 not-connected 100 32 disabled 0

2/40 1 not-connected 100 32 disabled 0

2/41 1 not-connected 100 32 disabled 0

2/42 1 not-connected 100 32 disabled 0

2/43 1 not-connected 100 32 disabled 0

2/44 1 not-connected 100 32 disabled 0

2/45 1 not-connected 100 32 disabled 0

2/46 1 not-connected 100 32 disabled 0

2/47 1 not-connected 100 32 disabled 0

2/48 1 not-connected 19 32 disabled 0

2/49 1 not-connected 4 32 disabled 0

2/50 1 not-connected 4 32 disabled 0

Does the port channel configuration is wrong?

Version 6.3(10)

show run

....

#port channel

set port channel 2/1-4 39

set port channel 2/5-8 40

set port channel 2/9-12 41

set port channel 2/13-16 42

set port channel 2/17-20 43

set port channel 2/21-24 44

set port channel 2/25-28 45

set port channel 2/29-32 46

set port channel 2/33-36 47

set port channel 2/37-40 48

set port channel 2/41-44 49

set port channel 2/45-48 50

set port channel 2/49-50 51

....

port1--Lucent CBX500(100Full)

port2--AS5300(100Full)

port3--AS5300(100Full)

port4--not connected

port5--other switch(10,Full)

port6--not connected

port7--not connected

port8--not connected

[milan.kulik]

Hi Patrick,

1) I was a little confused by your topology description

"The topo is:CBX500(Lucent switch) connect to port 2/1 of Cat2948,another switch connect to port 2/5 of cat2948."

I'm not a Lucent expert so I understood you are using a multivendor switching LAN with some Ethernet switching card in CBX500.

Is my understanding correct now that you are using just one Ethernet port to manage your CBX? And 00:60:d1:01:5b:f8 is the MAC address of the CBX management port directly connected to 2984 port 2/1?

What is connected to port 2/5? Another CBX management port? What is the MAC address of it?

Is there any other Ethernet port used on the CBX?

If not, I can't imagine how could the MAC address 00:60:d1:01:5b:f8 come to port 2/5.

2) Regarding the set port channel comands you provided - they only define channel port groups (I think they are default in 2948 CatOS config). To see the current channel config use "sh port channel" command or display complete module 2 config (sh config 2). It's always safer to switch off the channelling by "set port channel 2/1-48 mode off" if you don't use etherchannels at all in your network (it's saving some CPU load, I think). But be carefull - this command resets all the affected ports.

Regards,

Milan

lapping between port 2/5 and port 2/1.

I checked the physical connection,no loop,and STP enabled in cat2948.Mac address,00:60:d1:01:5b:f8,is the addr of CBX500.

When flapping happened,3 to 10 packets lost between cat2948 and CBX,this serious affectted our applications.I have refered the tech files

about Cat4000 serial switch error message,useless.

[patrick.zhang]

Hi,

Thank you for help.

My topo is :

switch(Lucent CBX500)----port2/1[Cat2948]port2/5----[unknown switch]----fa0/4[cat2924]---other devices.

Yes,only one port of cat2948 direct tconnect to CBX,mac 00:60:d1:01:5b:f8 should be the mac of CBX500.And show cam dyn is:

(enable) sh cam dy 2/1

* = Static Entry. + = Permanent Entry. # = System Entry. R = Router Entry.

X = Port Security Entry $ = Dot1x Security Entry

VLAN Dest MAC/Route Des [CoS] Destination Ports or VCs / [Protocol Type]

---- ------------------ ----- -------------------------------------------

1 00-60-d1-01-5b-f8 2/1 [ALL]

Total Matching CAM Entries Displayed =1

hapy-c2948-1> (enable) sh cam dy 2/5

* = Static Entry. + = Permanent Entry. # = System Entry. R = Router Entry.

X = Port Security Entry $ = Dot1x Security Entry

VLAN Dest MAC/Route Des [CoS] Destination Ports or VCs / [Protocol Type]

---- ------------------ ----- -------------------------------------------

1 00-03-a0-b3-f8-1c 2/5 [ALL]

1 00-50-ba-09-0f-8b 2/5 [ALL]

1 00-50-da-8f-3c-a2 2/5 [ALL]

1 00-80-2d-ec-bc-05 2/5 [ALL]

1 00-80-c0-10-0d-51 2/5 [ALL]

1 08-00-20-d2-21-09 2/5 [ALL]

Total Matching CAM Entries Displayed =6

[milan.kulik]

OK, so if the only ethernet port on CBX is connected to 2948, the only way how it's MAC address could appear on port 2/5 was some kind of network loop definitely.

I'd check the "unknown" switch.

What type and vendor is it?

Is there STP running on it?

Are any logs or SNMP traps available?

Are there some hubs connected to it with "too active" users who could make a loop on it?

An example what could have happened:

There was a hub connected to to the "unknown" switch. The hub was "intellingent" - it detected automatically the connected device type and was able to "cross" it's port pinout if another switch or hub was connected to it.

Some user connected a cable from a hub port to another port on the same hub.

CBX sent a broadcast to the network (ARP, e.g.).

This frame came to the "unknown" switch, to the hub and through the loop back to the hub, "unknown" switch and 2948 port 2/5 finally.

MAC address 00:60:d1:01:5b:f8 flapped from 2/1 to 2/5.

The user noticed "something is wrong" and disconnected the cable.

Problem "disappeared". There was nothing in any log except of the MAC flapping.

This story happened in my network a year ago.

I fixed this problem by spanning-tree portfast bpduguard on my access ports.

So your problem could be something similar. (The scenario decribed above would work even on any switch on your network if bpduguard is not enabled, not only on the "unknown" switch.)

Regards,

Milan

[patrick.zhang]

Hi,

thank you for help,it's a very good suggestion for me!

I can understand this story,so I'll check every host in my network for loop.

And I want to do another test that to set port cam filter on cat2948 port 2/5 which connect to unknown switch,the filter will filter packets which source address is mac of 00:60:d1:01:5b:f8,if the port 2/5 receive these frames,it discard them,so I think the port will not learn mac of 00:60:d1:01:5b:f8 again,so no flapping happen again.

But I'm not sure that port 2/5 will not learn the source mac addr before discar the frame,maybe the port 2/5 do the two operation,learning and discard.