>This is my inetd.conf!
>
>#inetd.conf of saturn
># amanda backup server with indexing capabilities
> amandaidx stream tcp nowait root /usr/lib/amanda/amindexd
>amindexd
I assume this is really just one line, right?
I don't see anything obviously wrong with this, so next we'll have to
do something harder.
I don't know/remember what OS you're running, but it probably has a
program called truss or strace or something like that that runs a command
and reports all the system calls. First, find that program and test it
on something simple. For instance, on my machine:
/bin/truss -fo /tmp/sleep.truss /bin/sleep 4
The -f option follows any forks and the -o option writes the output to
/tmp/sleep.truss. Just cat that file and you'll get the idea of what
it logs.
Now, comment out the amandaidx line in your inetd.conf file and add this
in its place (continuing my example, adjust as needed to match what you
found above):
amandaidx stream tcp nowait root /bin/truss amindexd -fo /tmp/amindexd.truss
/usr/lib/amanda/amindexd
Note the addition of "amindexd" before "-fo". That's a requirement of
the way inetd works.
When you try amrecover again, it should fail the same way, but you'll
have a system call trace in /tmp/amindexd.truss. Send that to me, please.
>When all the daemons mentioned above are set to the user amanda, it also
>doesn't work.
You should be using "amanda", not "root". But that is not what is
causing this problem.
>(after init 1/2)
Huh? After you change inetd.conf, you should be sending it a HUP signal.
I don't know if changing run states with init will make it reread the
config file.
>jörn h.
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
