Trunks going to non-trunk and flapping

[jclj]

Last week I had an issue where users were having trouble connecting to devices, interruptions, etc. I noticed some host flapping on quite a few ports in our engineering lab which looks to be a loop. I'm trying to figure out where/how this started. The confusion lasted for a few hours and then was gone. Why would we have ports go to non trunks after a loop? Our trunks are setup statically with a "switchport mode trunk". I could see this if DTP was negotiating. Also, why so many different ports flapping? I have seen this on at least 4 switches.

7/18/2014 10:52:34 AM 10.20.2.7 Warning 1892: 001919:   Host 00:1B:78:BC:A2:C6 in vlan 112 is flapping between port Gi1/1 and port Gi4/11
7/18/2014 10:52:32 AM 10.20.2.7 Warning 1891: 001918:   Host 00:50:56:A7:14:6F in vlan 112 is flapping between port Gi1/1 and port Gi2/5
7/18/2014 10:51:56 AM 10.20.2.7 Warning 1890: 001917:   Host 00:12:79:D5:E3:B3 in vlan 112 is flapping between port Gi1/1 and port Gi3/21
7/18/2014 10:51:56 AM 10.20.2.7 Warning 1889: 001916:   Host 00:50:56:A8:4C:D7 in vlan 112 is flapping between port Gi1/1 and port Gi4/11
7/18/2014 10:51:45 AM 10.20.2.7 Warning 1882: 001909:   Host 00:50:56:8E:5F:60 in vlan 112 is flapping between port Gi2/5 and port Gi1/1
7/18/2014 10:51:45 AM 10.20.2.7 Warning 1881: 001908:   Host 98:4B:E1:75:E4:48 in vlan 112 is flapping between port Gi3/32 and port Gi1/1
7/18/2014 10:51:42 AM 10.20.2.7 Warning 1880: 001907:   Host 00:11:43:DC:0E:EB in vlan 112 is flapping between port Gi1/1 and port Gi2/7
7/18/2014 10:51:41 AM 10.20.2.7 Warning 1879: 001906:   Host 10:60:4B:A9:81:8C in vlan 112 is flapping between port Gi1/1 and port Gi3/32
7/18/2014 10:51:38 AM 10.20.2.7 Warning 1878: 001905:   Host 00:12:79:D5:E3:B3 in vlan 112 is flapping between port Gi1/1 and port Gi3/21
 

000737: Jul 18 10:52:38.076 MDT: %SW_MATM-4-MACFLAP_NOTIF: Host 0050.5649.6d62 in vlan 112 is flapping between port Gi0/38 and port Po10
000738: Jul 18 10:52:39.091 MDT: %SW_MATM-4-MACFLAP_NOTIF: Host 0011.0a57.9382 in vlan 112 is flapping between port Po10 and port Gi0/6
000739: Jul 18 10:52:44.125 MDT: %SW_MATM-4-MACFLAP_NOTIF: Host 00c0.dd0c.b551 in vlan 112 is flapping between port Gi0/40 and port Po10
000740: Jul 18 10:52:48.092 MDT: %SW_MATM-4-MACFLAP_NOTIF: Host 0050.5649.6d62 in vlan 112 is flapping between port Gi0/38 and port Po10
000741: Jul 18 10:52:52.270 MDT: %SW_MATM-4-MACFLAP_NOTIF: Host 0050.564e.4e8e in vlan 112 is flapping between port Gi0/37 and port Po10
000742: Jul 18 10:52:52.907 MDT: %SW_MATM-4-MACFLAP_NOTIF: Host 0050.5647.3916 in vlan 112 is flapping between port Gi0/11 and port Po10

7/18/2014 4:45:16 PM 10.20.2.7 Notice 2056: 002085:   Port Gi2/46 has become non-trunk
7/18/2014 1:50:49 PM 10.20.2.7 Notice 2032: 002061:   Port Gi2/1 has become non-trunk
7/18/2014 1:50:41 PM 10.20.2.7 Notice 2029: 002058:   Port Gi2/4 has become non-trunk
7/18/2014 1:50:14 PM 10.20.2.7 Notice 2026: 002055:   Port Gi2/2 has become non-trunk
7/18/2014 1:49:42 PM 10.20.2.7 Notice 2017: 002046:   Port Gi3/6 has become non-trunk
7/18/2014 1:49:32 PM 10.20.2.7 Notice 2014: 002043:   Port Gi3/4 has become non-trunk
 

Thanks

[Tom Watts]

Host flap can be a difficult symptom to resolve. It may be a cause from several different issues

- Do you have spanning tree enabled? If not, enable RSTP and ensure PORT FAST is not enabled on any switch interconnecting interface

- Do you have multiple gateways? It's possible a device is replicating a default gateway causing the source MAC to advertise 2 different way

-Do you have multiple DHCP pools from different sources? May be possible bleed over causing confusion

-We notice there are several ports, shut down every port then turn up one of the ports and see if the flap stops or not, this will help get you to find where a possible loop is at

-Are you load balancing over a LAG or something else, especially with servers? If load balancing with servers, make sure each NIC uses a different MAC or shut down one NIC for testing.