How to monitor the health of the MailScanner architecture

[Holger Nöfer]

Hi,

for the monitoring of my server I use hobbit.
It has very nice plugins.
If you use hobbit you have a hobbit-server (Monitoring server)
and a hobbit-client (e.g. Mailserver).
The Server does some checks from the server side an the
client reports some values, like cpu load, disk usage, inodes, mailq
items, mailq size and so on. On both sides, server and client, you can
create your own scripts which can monitor your system.

At the server side you can connect a gsm modem to the server to notify
an admin if the server hase some problems or you can use emails to
notify them.

It is very nice ;-)

Best regards,
Holger

Jonas A. Larsen schrieb:
> Hello all
> 
>  
> 
> I have a problem, and discussing it on the irc channel didn’t turn up
> any obvious solution.
> 
>  
> 
> Say you have more than 1 MS box scanning mails for a specific domain.
> They are load balanced in some way, so the load is split over the servers.
> 
>  
> 
> Now lets say one of the servers have a problem. Not a fatal problem, so
> the server is still running (responds to pings etc) port 25 is still
> open, and exim (the mta in my case) still accepts mails.
> 
>  
> 
> But for some reason, crash, corrupt config, full root fs etc. the
> process of moving mails from the incoming queue to the outgoing queue is
> not working.
> 
>  
> 
> What I am interested in, is a system to alert me of such a problem
> automatically.
> 
>  
> 
> Currently the only thing, besides clients noticing mail being delayed,
> is for me to look at my mailscaner-mrtg graphs for the incoming queue
> and notice that its growing.
> 
>  
> 
> One method of doing all this automatically that we came up with, would
> be some complex system that would work as follows:
> 
>  
> 
> You create a domain for each MailScanner, that only that MailScanner
> scans for.
> 
>  
> 
> You then create an imap account on another system for each of the domains.
> 
>  
> 
> You then create a script that sends a mail to each of the accounts and
> after X amount of minutes check to see if the mail has arrived on the
> imap account. If yes, delete the mail and do the same thing again after
> Y amount of minutes (a cron job), if it doesn’t exist something must be
> wrong with the mailflow, either its interrupted or is experiencing delays.
> 
>  
> 
> Do anybody have a better idea or know of something that can do this already?
> 
>  
> 
> My root file system ran full last week, and it caused mails to still be
> accepted (incoming is on /var on another disk) but MS was frozen because
> it couldn’t extract attachments to /tmp which was full because it was on
> the same disk as the root fs.
> 
>  
> 
> I hope I have made the above somewhat clear, if not please ask me to
> clarify.
> 
>  
> 
>  
> 
> *Med venlig hilsen / Best regards*
> 
>  
> 
> *Jonas Akrouh Larsen*
> 
> * *
> 
> TechBiz ApS
> 
> Laplandsgade 4, 2. sal
> 
> 2300 København S
> 
>  
> 
> Office: 7020 0979
> 
> Direct: 33369974
> 
> Fax:    7020 0978
> 
> Mobile: 51201096
> 
> Web: www.techbiz.dk <http://www.techbiz.dk/>
> 
>  
>