🚀 go-pugleaf

We have two seemlingly identical installations of visibroker 3.2 on
two different unix workstations. The two workstations are on the same
NIS-network so the environments are very similar.

The problem is that one of the installations seems to think that it is
on a machine named 'localhost'. The other machine works fine and seems
to think (correctly) that it is running on a machine named
'foo.ericsson.se'. What I mean by "seems to think" is that when I run
'osfind' on a third machine, the list of implementations shows
'HOST:localhost' for implementations on the former machine and
'HOST:foo.ericsson.se' for implementations on the latter machine.

Furthermore, server processes running on the 'localhost' machine
cannot be contacted by clients running on the third machine, while
server processes running on 'foo.ericsson.se' can be contacted just
fine by clients running on the third machine.

We have really made an effort in making the two environments as
similar as possible. All agents and server processes are run by the
same nis-registered user with identical environment variable settings
in all sessions.

Does anyone recognize this? How does osagent find out what machine it
is running on? Can I force osagent to believe something different?

-- 
   Klas Mellbourn
   finger for public key
   http://www.al.etx.ericsson.se/~qtxklme