We recently ran into an issue on VEM Version: 4.2.1.1.4.0.0-2.0.1 that began as a N5K vPC / port-channel issue. We were troubleshooting why a N5K peer link died, and from an etheranalyzer capture, we saw a huge amount of LACP PDUs on the control plane coming from a MAC assigned to a VEM. Upon further investigation with Cisco, we learned the vemdpa process went buggy and was sending out lacp pdus at a rate of several hundred per second, this LACP flood from the VEM was over-running the LACP process on our N5K's which killed the vPC peer link between the N5K and the upstream facing vPC to the N7K. . The catalyst of the event maybe a VSM supervisor failover.
These is the versions on the rest of the environment:
VEM Version: 4.2.1.1.4.0.0-2.0.1
VSM Version: 4.2(1)SV1(4)
Workaround is to kill vemdpa process on the host. Restart.
This is an undocumented VEM issue at this point.
No comments:
Post a Comment