ETHCNTR-3-LOOP_BACK_DETECTED on switches singly connected to the distribution switch

[sgonsalv]

Hi Guys,

Come across a strange issue where we are seeing the following error:

"%ETHCNTR-3-LOOP_BACK_DETECTED:

on all uplink ports connected to downstream switches from the distribution switch.  These downstream switches are singly connected by one uplink, so there isn't a "loop" as such.

There haven't been any changes made to the network and only this morning did we notice that all downstream switches UTP uplinked to the distribution switch was down in an err-disable state.  Based on reading some of the documentation there is mention of replacing cabling, which we've done to no avail.  As a temporary workaround we've disabled keepalives to restore network connectivity, however i'm wondering what would be causing the downstream switch to see the keepalives that it sends on the uplinks back to it??

The switches are running 12.2(55)SE5.

Also, i've checked spanning tree for all VLANs on the distribution switch, which is dual uplinked back to the core, and for each VLAN there is only one path to the root, since the other interface is blocked, so there doesn't seem to be a spanning tree loop as such.

Any asistance would be appreciated.

[InayathUlla Sharieff]

HI,

The reason for the ETHCNTR-3-LOOP_BACK_DETECTED message is that we are

receiving our own ethernet keepalive back. Believe the problem to be with the GIG interface receiving it's own keepalive packet.

Thus it senses a loop and errdisables the port.

Thus I would ask you for the following:

- have you tried swap gbics/cables?

There are couple of bugs filed on this issue. check out this

http://www.cisco.com/cgi-bin/Support/Bugtool/onebug.pl?bugid=CSCea46385

Another one is CSCdz72393

There are numerous bugs filed against this issue and the request to have keepalives

disabled by default on fiber ports.

To avoid such problem we entered command on the gig interface

"no keepalive"

You can also recover errdisabled ports by entering command

errdisable recovery cause

Please note that keepalives can be safely removed and the interfaces will still remain up

on the link partner. Beside, also note that the keepalive is not used to determine the

interface status like on Router Ethernet ports. Consequently the port will go up or down

irrespective of the keepalive setting.

DSBU switches (2950, 3550, 3750) send keepalives out all interfaces by default. This was

implemented since the first release to prevent loops when customers use old Type 1 cables.

The use for keepalives on switches and routers are different, and I do believe that there

is an argument

that could be made that keepalives should be disabled on fiber ports that cannot use Type

1 cables.

HTH

Regards

Inayath

*Plz dont forget to rate if this info is usefull.

[sgonsalv]

Hi There,

Thanks for response.  Yes we have replaced cabling and some of the UTP SFPs to no avail.  I can't seem to get any more information on the bug IDs you've mentioned - can you provide me with more information?  Are these bugs seen for the verison of code i've mentioned: 12.2(55)SE5?

The other point worth making is that we are seeing these loop back detected errors on the UTP connected ports..the only fiber uplinked ports are from the distribution switch to the core - these are not affected. Hence why we've disabled keepalives on these UTP uplinks. However i'm curious as to why the recommendation to disable keepalives is only talked about wrt to fibre uplinked ports and not UTP?

thanks in advance

Sheldon

[canero]

hi,

We are experiencing a similar issue with 2960S switches ( WS-C2960S-48LPD-L) with IOS c2960s-universalk9-mz.150-2.SE2.bin. When we check the "show interface gi1/0/49" we see that keepalive is enabled, but since this is a fairly new version we don't think there can be similar bug. I think if keepalives can create a similar problem then by default it should be disabled on 150-2.SE2 train but we see that they are enabled. Can there be any cases related to Spanning-Tree version or any other necessities where keepalives should not be enabled?

http://www.cisco.com/en/US/tech/tk389/tk621/technologies_tech_note09186a00806cd87b.shtml

Keepalives are sent on all interfaces by default in Cisco IOS Software Release 12.1EA-based software. In Cisco IOS Software Release 12.2SE-based software and later, keepalives are not sent by default on fiber and uplink interfaces. For more information, refer to Cisco bug ID CSCea46385 (registered customers only) .

The suggested workaround is to disable keepalives and upgrade to Cisco IOS Software Release 12.2SE or later.

The recommendation is not to disable to error disable recovery time, but solve the root cause of the problem. Is there any recommendation to find out where and on which interface this problem first began.

best regards