How to monitor mail servers

I’ve acquired a new piece of server monitoring software (I refuse to pay $40/month PER URL to some monitoring service), and I’m trying to figure out how to monitor DH’s mail servers.

With the problem that we just had today on Hoggle, normal monitoring didn’t show a glitch, apparently since the server is up-- just not coughing up the mail.

So, when I just monitor port 110 (POP3), the server looks responsive, even though it’s not.

And monitoring port 220 (IMAP) doesn’t work at all with DH; the software sez the servers are always down. I take it this means DH is somehow using port 110 for both POP3 and IMAP…? Or you’re using another port for IMAP…?

It appears the only way to truly know if the servers are working is to check them and test for a certain string in response. The software (“Server Sentinel”, Mac) allows me to enter a server name, a port number, and some optional string to test for. So, the question is, what string do I want to test POP3 for?

And do I need to test separately for POP3 and IMAP? If so, are there different strings to test for?



telnet 110 will check for POP (143 for IMAP; 25 for sendmail)

  • if you get a sucessful login, it will look like this:

jazz % telnet hoggle 110
Connected to
Escape character is ‘^]’.
OK Cubic Circle’s v1.31 1998/05/13 POP3 ready c7540000e0a40d3b@hoggle

This is useful for debugging problems with POP. If you want to test an actual string, you’ll need to do something like:

USER username
[it responds]
OK username selected
PASS password
OK Congratulations!

LIST shows the messages; RETR n retrieves message number ‘n’.

If you want to check incoming mail, check for a connection to port 25.
Right now (cough) it’s not working:
jazz % telnet hoggle 25
telnet: connect to address Connection refused
telnet: Unable to connect to remote host

after restarting:
jazz % telnet hoggle 25
Connected to
Escape character is ‘^]’.
220 ESMTP Debian Sendmail 8.12.0.Beta5/8.12.0.Beta5/Debian 8.12.0-1; Thu, 24 May 2001 17:20:33 -0700

Basically sendmail stops itself if the load gets too high. We are in the process of replacing the temporary filer on that cluster with a new filer (Network Appliances) which will resolve almost all of the mail problems on this machine. In the mean time we’re moving the mail spool to a local drive which should make things better on a more temporary basis.

Hi, Will…

Thanks for the info…

Unfortunately, I can’t really use it, since the kind of monitoring I’m talking about is automatic, not interactive, and I apparently can’t script this software with prompts and responses.

The real problem occurs when sendmail quits, leaving the server deceptively running. I guess I can either just hope that the problem really is going to be fixed, find a different way to monitor the server, or convince the software author to add some scripting.

Thanks again…



I just realized that the version of your post that I replied to – the one that was emailed to me – was different from the final post… which you apprently edited. Looks like wwwthreads only sends a single email for a post, and doesn;t remail after editing.

Anyway, the strings in your post didn’t work for me… But after my previous post, I decided to telnet into my server and see what I could see.

It turns out that the strings returned by the server that I was testing are different from what you posted; I was trying to test for “Debian Sendmail” – but Dot returns “ESMTP Sendmail”… with the “Debian” coming later in the version line.

So, it looks like I got it working after all.

Thanks again!