A node issue could affect to entire Virtual server with 5 nodes?

Three if's:

1. if the bad pool member was still passing its assigned F5 monitor
2. and if the other pool member(s) had existing connections
3. and if the nature of the failure was for the pool member to prematurely terminate connections (leading to a lower current connection count for the bad node)

Then the F5 would send most-if-not-all traffic to the bad node