Re: Probleme mit Sendmail
Richard A Nelson wrote:
> I see two *BIG* problems with your setup that you'll need to rectify
> before we can do much about your sendmail errors:
> 1) Exporting /var
> Have you considered the ramifications of this ? What happens
> when a client boots up and tries to clean /var/run ? or when
> a daemon starts on one machine, then tries on the other ?
> I can see exporting /var/mail/ (or /var/spool/mail), but
> am not sure what rational advise I can offer in this regards.
This is not an issue on my system. I use NFS to shut up the noisy hard
disk in the system on my desk. My home directory is located on the
server and exported via Samba, so I can access it from the Windows
machines in our network. When I realized how nice and quiet my machine
turned, I also moved /var and /usr, to shut it up for good. I'm the
only one using Unix, and as you can see, all the NFS exports are
exported to only one host: bart, which is my desk machine. This is also
the reason, why each exports has the no_root_squash options set. It's
simply not an issue with my system and much more convinient for me.
Other than that, you're quite right.
> 2) Unable to lock
> If locking isn't working, you're going to soon be hosed -
> any attempts to simultaneously access a file may leave it
> in a inconsistant state.
I misquoted the error message:
lockd: cannot monitor 192.168.0.1
lockd: failed to monitor 192.168.0.1
These repeat for a screenfull on bootup and then just stop. On both the
server and the client, lsmod shows a module called lockd, so I figured,
that this error message isn't really a concern for me. The kernel wants
to do something that requires lockd, lockd gets loaded by modprobe while
the kernel is bitching about lockd being not ready, once lockd is
successfully loaded, the kernel shuts up. This is just my idea of the
world, I have not verified this.
What about sendmail?