Chain of events in the case that was reported to Check Point:
- Cluster members are connected to a switch
- The switch is rebooted
- The switc gs the physical link "UP" on its ports before it is ready to pass traffic
- Since the link is "UP" on the switch's ports, the link is also goes "UP" on the slaves of the Bond interface on cluster members
- The Bond interface on the cluster member goes in to "Active" state
- If the option "Switch to higher priority Cluster Member" is configured in the cluster object, the cluster fails back to the higher priority member for a few seconds
- This causes a short traffic outage