Dear Cisco community,
I had an issue with an access switch (C2960X) that has been disconnected from the whole network because its uplink went into an "err-disabled" state. The access switch has only one L2 copper uplink to one of our core switch through an access port in a specific vlan (no trunk mode).
The issue has been solved by shut/no shut the port and has not happened again since then.
Here are the logs :
Oct 1 14:44:42 UTC: %ETHCNTR-3-LOOP_BACK_DETECTED: Loop-back detected on GigabitEthernet4/0/48.
Oct 1 14:44:42 UTC: %PM-4-ERR_DISABLE: loopback error detected on Gi4/0/48, putting Gi4/0/48 in err-disable state
Oct 1 14:44:43 UTC: %LINEPROTO-5-UPDOWN: Line protocol on Interface GigabitEthernet4/0/48, changed state to down
Oct 1 14:44:44 UTC: %LINK-3-UPDOWN: Interface GigabitEthernet4/0/48, changed state to down
Based on my researches, it seems that the issue may come from a keepalive packet that went back through this interface due to a loop. I'm concerned because even if the "no keepalive" avoid a future "err-disabled" state, it won't correct anything regarding an eventual loop.
However, I have no clue at all that there is a loop somewhere in the network. The STP has been confirmed to be correctly configured and working on every switch. Also, after shut/no shut the port, everything went back normally.
- What could cause a keepalive packet to comes back only once though its initial interface ?
- How could I be sure that there's no loop in our network ? The network global performances are actually pretty good and I have no visible sign that it exists an issue of this kind.
It may be important to note that :
- We're using HSRP between our core switches.
- I have seen that portfast was enabled on every side of the switch uplink to our core switch and know that this is not recommended. But I still don't get how a potential loop could have been formed. In the meantime, I have removed this setting.
I'm counting on your experience to help me. ;)
Thanks a lot !
