>I took amanda from the RPMs, so I don't know any internals besides the
>config-files.
And therein lies a problem when you try to get help.
>I'm a bit afraid of compiling the program myself, because I remember my bad
>experiences in compilng the 2.4 Kernel ...
Building Amanda should be **nothing** like trying to build a kernel.
>Recently, I downloaded lsof and grepped its output for am*.
Good plan.
>As You can see, amandaidx and imidxtape are listening, but there is no
>amandad:
So we need to get that fixed first.
Use of xinetd has been on the mailing list often recently. More or less
at random, I found this entry, but note that I have not run it myself:
service amanda
{
protocol = udp
socket_type = dgram
wait = yes
user = backup
group = disk
server = /usr/local/libexec/amandad
disable = no
}
Obviously you'll need to change the path to amandad and the user and
group.
>The compiled-in user "operator" HAS write-access to /tmp/amanda.
That was not what I asked. I asked if "operator" has read/write access
to its own home directory. Silly question but it has bitten more than
one person from the RPM's lately.
>When I start amandad by hand, it waits 30 secs and prints a Timeout
>message to /tmp/amanda/amandad.debug.
That's normal. Did you run it as root or operator?
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
