Thomas,

We recently had a similar problem on a high availability system application
running on a Solaris platform with multihomed nics.  We are running Netview
V5.1.3 on a Solaris box as well.  In our case Netview reported the private
Network down, however when our Unix Admins checked the interfaces on the
Servers there were no indications of link loss or collisions.  All traffic
was passing EXCEPT for ICMP.  Our switches (will not reveal the vendor name)
had a problem with their management module falling asleep , therefore it
just drops any packets not deemed important.  ICMP packets are at the lowest
priority..   

Why your V3 box did not see the bounces??  Maybe there was a change in the
priority of the ping in V5??  When switches get real busy, pings are
basically the lowest in the food chain.... 

In our case Netview was doing its job correctly, and it fact made us more
proactive in replacing our switch infrastructure to avoid having a complete
hardware failure.

Just another situation, not sure this is your problem, but I would verify
your switches are passing the management packets.