How to monitor the health of the MailScanner architecture

[Ken A]

Jonas A. Larsen wrote:
> Hi Glenn
>> Um, just script up a snippet that send a mail through to a service
>> account (use telnet with expect, or perl or whatever... Make that
>> sending snippet a function/sub/procedure and let it take an argument
>> servername, then loop through the list of servers.... You know what to
>> do:-), then use an automated MUA in the same script (whatever you
>> need) to check that it is received within a reasonable time... Nail
>> within an expect script would be nice for that last bit ... How hard
>> can it be:-):-).
>> You'd have to keep an eye on reasonable timing, and think through how
>> to report errors...
>> Could probably be incorporated as a testscript into any monitoring
>> app... Or run from cron with some reasonable regularity.
>> Should be fairly easy to write up ... But I'm on vacation, so you do
>> it yourself;-).
>> This is precisely what is needed :) regarding the "fairly easy to write
> up" I guess that depends on how elite coding scriptiong skills you have and
> how much time you got :)
> 
> We currently got it running using a freeware windows monitoring tool, that
> had this precise check.
> 
> I was just looking for something that already existed and general comments.
> (because I'm a lazy boy)
> 
> People seem to have misunderstood it a bit since they are recommending
> generic monitoring solutions. Neither nagios, bigbrother or others have the
> above solution. I actually think munin had something but it would have to be
> re-scripted to be useable (assuming I don’t want to run munin)
> 
> I do find it odd though, that more or less nobody appears to be monitoring
> their mail systems in this way. I still say it’s the only way to be 100% if
> your system is functioning or not. Checking the mta daemon, mailscanner
> daemon, queue sizes etc. are all not a perfect way to check if the
> mailscanning process is functioning.

There are many parts to the mail flow process. You can't check them all 
by passing a message through, though that certainly will tell you that 
messages can flow through your mail system! If I get a page at 3am that 
says "mail stopped flowing", I'd be pretty disappointed in my monitoring 
software. I want to know a lot more than that before my feet hit the floor.

If your monitoring software is setup correctly (nagios here), you can 
quite correctly infer that all processing is working normally if all 
tests are passed. You just have to design or tweak the tests (these are 
usually simple shell or perl scripts that nagios uses) to fit your 
architecture. It's not hard, but it's implementation specific, mostly.

Ken

> 
> Cheers
> 
> Med venlig hilsen / Best regards
>  
> Jonas Akrouh Larsen
> 
> 


-- 
Ken Anderson
Pacific.Net