I read that article and I see workaround CP has, but not sure if thats something most clients would do, to switch from unicast to broadcast. Even in HA active-standby, CCP mode by default is ALWAYS unicast and works fine. I worked with customer before who also had Cisco Nexus (not sure exact model number), but they never had this issue, though they had cluster XL HA, not load-sharing.
Here are some questions I have:
1) Were you guys always using same switch?
2) If answer to 1 is yes, when did this issue happen originally?
3) If answer to 1 is no, what was behavior previously?
4) How often does the problem happen?