UNIX for Dummies Questions & Answers

Hi there,
I am having a small issue with the mail function on our controllers. Recently we set up all the boxes as NFS slave servers and mail sending was not affected. We then had to change the servers addresses in resolv.conf and now email is being queued and not being sent.

I have restarted the sendmail on the machines and the issue still is there. What could be the issue? We have other controllers with the resolv.conf and the mail function does not seemed to be affected - the only difference is that they are clients not slaves.

I have changed back to the old ip addresses in the resolv.conf and mail sends fine but these will be going offline very soon..I know this give me a clue that it is these addreses that are causing the conflict.

I have also compared the mailsevrs files and they are also matched.
What else could I check? I have looked at the fstab comparing them with other controllers and they are matched.

Rgds

Lodey

Sendmail requires DNS to work, does 'nslookup some_domain_name.com' work now that you have changed resolv.conf
Does traceroute to the dns server work?

Depending on the version of sendmail and sendmail.cf used where /var/mail is mounted from can determine what sendmail decides is the mail server, so changing /var/mail from being a local filesystem to a NFS mounted one could affect sendmail.

If your system is using DNS to determine where to send mail to then you can check it by running nslookup and querying it as follows (you type in what is in bold):

Code:

$ nslookup 
> set type=mx 
> mailhost 
Server:  abc.abc.domain.co.uk 
Address:  192.161.191.3
mailhost.abc.domain.co.uk        preference = 10, mail exchanger = server.abc.domain.co.uk 
abc.domain.co.uk nameserver = abc.abc.domain.co.uk 
abc.abc.domain.co.uk     internet address = 192.161.191.3 
>

Compare what you get from the new DNS servers and the original ones

Another thought,
You might add a smart host in the sendmail configuration that points to a known working sendmail system.