kprop problem (slave installation)

[Ken Raeburn]

On May 19, 2004, at 10:12, Domenico Diacono wrote:
> On the slave the only log that has a trace of the failed transaction 
> is the
> auth.log:
>
> May 19 15:30:11 slave xinetd[1237]: START: krb_prop pid=1454 
> from=ip.master
> May 19 15:30:11 slave xinetd[1454]: FAIL: krb_prop address 
> from=ip.master

Sounds like the server side is dying right away.  Check some basics 
first:  Is the pathname in the xinetd config file correct?  Can the 
specified user (probably root) run the executable?  Is it the right 
executable?  Does it crash immediately if you try running it from the 
command line?  Are there any core files in the directory that xinetd 
runs programs in?  Does it complain about the command line arguments if 
you run it with those specified in the xinetd config file?  Did you 
give it the "-S" argument that tells it that it, and not xinetd, should 
be handling new incoming connections?

Check your syslog configuration to see where daemon.info (and more 
severe) messages will be logged, since that's what kpropd uses; if 
you're not logging them, you should change that.

Worst case, you could set up xinetd to run kpropd under strace (for 
Linux; truss for Solaris; ktrace for NetBSD; etc) and log the syscall 
activity someplace where you can examine what's going on.  (Don't post 
the output, it'll probably include private key data.)

Ken