No Layer 2 communication between terminals after M-LAG member port failure

2023-12-13 17:19:11 Published
  • 0 Followed
  • 0Collected ,727Browsed

Network Topology

Device and version: S6890-54HF R2820

Networking: Two S6890 devices are configured as M-LAG leaf devices on-site, and two downstream servers are interconnected through M-LAG interfaces. The following diagram shows a partial network diagram:


Problem Description

Two S6890 devices are configured as M-LAG leaf devices on-site, and two servers in the same network segment are connected. A link failure occurs on port 1/0/9 of TOR2, which leads to communication failure between S2 and S1. The ARP and MAC entries of the terminals are normal on both leaf devices and are learned on M-LAG interfaces.


Process Analysis

After the link failure between TOR2 and server S1 on port 1/0/9, the traffic from server S1 to server S2 can only pass through the peer-link link. At this time, S1 and S2 cannot communicate with each other, so we need to focus on the peer-link link.



Checking the configuration of the peer-link link, we find that the often overlooked static source MAC check is also configured correctly.


 

The Bridge-Aggregation100 member ports of the peer-link link are also selected correctly.



Further checking the underlying status of the physical member ports of the aggregate 100, we find that one of the member ports, HGE1/0/30, is in STP block state. When packets are hashed to this port, they cannot be forwarded normally.




The peer-link interface does not participate in STP calculation. It is abnormal for the member ports of the peer-link to be in STP block state. Therefore, it can be basically determined that this is not a configuration or networking problem.

After subsequent positioning, it is confirmed that this is a software issue in the current version. After enabling stp global enable globally, the aggregated member ports may be in STP block state due to oscillation. This issue can be resolved by installing a patch.

Solution

The issue can be resolved by installing the R2820H01 patch. For environments where the issue has already occurred, the interfaces need to be shut down/undo shutdown to be restored after the patch is installed.

Please rate this case:   
0 Comments

No Comments

Add Comments: