Fernando,
Seems that the issue persists in Eth1/36 as well based on the event-histories you shared. Here's what the N3K logs when the port bounces:
Background information:
The debounce logic is written in NX-OS to detect when we lose signal from our directly connected host or device. When signal is lost, the Nexus will start a debounce timer which is 100ms by default to wait for the other end-device to signal us back, if we do not detect anything at this point we will take the link down assuming the connected port is down as well to avoid black holing traffic. This can be modified up to 5 seconds but it is against best practices to do so. This can be done at the interface configuration level with "link debounce time 5000”. In other words, based on the Nexus logs, it would appear that the issue lies on the other side of this connection which could also explain why the problem is still present in Eth1/36.
Sample output from the file you shared:
'show platform software ethpc event-history interface e1/36'
<snip>
33) Event IF_PCFG_CMD, len: 8, at 444263 usecs after Sat Oct 21 05:47:47 2017
Processed port command PORT_CMD_DISABLE(3) status 0
34) Event IF_LINK_ST, len: 34, at 436138 usecs after Sat Oct 21 05:47:47 2017
Link down sent to 0xfe000000 - Failure, Reason: Link down debounce timer stopped and link is down
35) Event IF_LINK_ST, len: 34, at 435893 usecs after Sat Oct 21 05:47:47 2017
Link down sent to 0xfe000000 - Failure, Reason: Link down debounce timer started
36) Event IF_LINK_ST, len: 34, at 268110 usecs after Sat Oct 21 04:51:22 2017
Link up sent to 0xfe000000 - Success, Reason: SUCCESS, Speed: 10000, Mdix mode: 3, Flow control mode: 0, Port duplex: 2
<snip...another example...>
59) Event IF_LINK_ST, len: 34, at 554438 usecs after Sat Oct 21 04:50:12 2017
Link down sent to 0xfe000000 - Failure, Reason: Link down debounce timer stopped and link is down
60) Event IF_LINK_ST, len: 34, at 552828 usecs after Sat Oct 21 04:50:12 2017
Link down sent to 0xfe000000 - Failure, Reason: Link down debounce timer started
61) Event IF_LINK_ST, len: 34, at 916406 usecs after Sat Oct 21 04:45:54 2017
Link up sent to 0xfe000000 - Success, Reason: SUCCESS, Speed: 10000, Mdix mode: 3, Flow control mode: 0, Port duplex: 2
I'd contact the carrier that owns the router/end-device and have them troubleshoot L1 at their end. Naturally, if there's a patch on your side, try a different patch panel port to be sure as well.
Hope that helps!
- Andrea
- Andrea, CCIE #56739 R&S
