The below described problem has been resolved: 

Some small errors in the NIS setup caused the problems (mainly in /etc/hosts) 
caused confusion in NIS)

Gerrit Hommersom 

===============================
All

I have changed the NIS server to a new system (including a NIS domain name 
change) that is currently not in the amanda back-up scheme. The amanda host and 
one client have moved to the new NIS domain.

The effect is that I get 'selfcheck time out' messages of the computers moved 
to the new NIS domain: the amanda host and ClientA. The two clientsB and 
clientsc  remaining in the old NIS still work proper. amanda check debug files 
generates two "no feature set" messages from the clients[B,C] that do backup. 
No messages related to the failing host and client A nodes. The messages  file 
from the amanda host contains: 
     xinetd: getpwuid(32783)(service amanda) failed: no child process) I use 
amanda.2.4.5

The amanda backup server was not and is not the NIS server. I found the 
reference of Florian Lengyel (58148) on his issues with NIS server merger: but 
that does not seem to apply here. 
Password/group files have been copied from the previous server so amanda 
UID/GID are the same. 
I can manipulate files under the new NIS server regime as "amanda".

The NIS server:
  has amanda services in /etc/services file on the correct UDP port. 
  no amanda daemon is running
  firewall settings allow free flow of commands between the amanda host/client.

Questions:
What components of amanda do I need to install (minimum) on the NIS server to 
make it work. What else may have gone wrong i.e.where do I need to look. Only 
amcheck debug appears in /tmp/amanda.

Many thanks
Gerrit Hommersom