a) time outs
rpc.acedbd is declared wait in inetd.conf
hence 1) it is forked only if it does not yet exists
2) it is not killed by unix at end of connection
however, aceserver commits suicide after its own time out
which is a parameter on the command line here 600 seconds
since the decalration is:
wwwace2/1 stream rpc/tcp wait taka /usr/local/bin/rpc.acedbd rpc.acedbd /usr/people/taka/wwwace2 20000200 600:600:0
***********************
b) declarations are required i think in 3 places: /etc/inetd.conf, /etc/rpc /etc/services
these are the entries on my own machine:
/etc/services
# These are acedb databases servers
cele 20000100/tcp rpc.acedbd
igd_ace 20000200/tcp rpc.acedbd
# These are acedb/jade netclients
worm 20100/tcp #aceclient
/etc/rpc
#acedb worm data server
cele 20000100 rpc.acedbd
igd_ace 20000200 rpc.acedbd
/etc/inetd.conf
# experimental ACE client to use with java scripts
worm stream tcp nowait mieg /usr/local/bin/netclient netclient -host beta -port 20000100
cele/1 stream rpc/tcp wait mieg /usr/local/bin/rpc.acedbd rpc.acedbd /local1/worm 20000100 1200:1200:0
igd_ace/1 stream rpc/tcp wait mieg /usr/local/bin/rpc.acedbd rpc.acedbd /local1/igd/ace 20000200 1200:1200:0
I suspect that you may be missing the entry in /etc/services
in particular I think that it is crucial in services to associate the
service to its port and to its executable (here rpc.acedbd)
I beleive the system runs corectly on sgi, but i have myself
set it up only on a dec alpha (with the precious help and
collabortion Bigwood & co) and on a Solaris with Donn Davy)
If this system does not work, please let me know.
***********************
c) rebooting
In principle, one should not reboot, just restart the correct daemons, but
in my experience, i always end up rebooting, becasause after killing
and restarting a few daemons, acedbserver may well run, but some other
service like nfs or whatelse is unhappy.