On Wed, 12 Nov 1997 Bruno.Vuillemin@??? wrote:
> The idea is for HP-Openview to get alerts when exim encounters some kind
> of problem, for example : a local host that doesn't accept any new mail
> and the queue become bigger and bigger on exim.
>
> (proposition 1): The principle is that exim (internally) should trigger
> snmp traps and send them to HP-Openview, which in turn would process them
> and eventually send an alarm to our pagers.
>
> (proposition 2): An other possibility is that exim, under certain
> circumstances, runs a script that raises the snmp trap.
A third possibility is to write an entirely separate program that works
the way the Exim Monitor works - by reading Exim's log file continuously.
This avoids any problems of interlocking that would otherwise have to be
solved because Exim does not have a central controlling process.
On Wed, 12 Nov 1997, Richard Welty wrote:
> we're thinking about parking Linux mail relays and
> DNS servers with Exim and Bind in our POPs, and keeping track of all
> of them is going to be pretty miserable.
Currently, an X screen full of Eximon displays is the best you can do, I
think. (I normally have 3 production ones on my screen, plus some
testing. When the manager of our other systems was on holiday, I had
five or six for a time.) Note that you can get them to be coloured
differently by starting them with commands like "eximon -bg blue".
On 12 Nov 1997, Stuart Lynne wrote:
> But for smaller sites that don't have an SNMP infrastructure how
> about some way of getting the running daemon to dump stat's like
> named does (either on demand or periodically).
The problem with that is that the running daemon doesn't have any stats!
Also, you can run an Exim host without having a daemon at all (input via
inetd; queue runs via cron). The "no central coordination" design of
Exim makes it lightweight, but means the only overall repository of
stats information is the log.
> The type of information I'm most interested in seeing are requests
> per minute for the last minute, five minutes, fifteen minutes (mail
> load average); CPU usage for parent and children since start and
> current; total messages processed (delivered, failed, rejected,
> queued, frozen, etc).
A log-scanning program could handle that, except for the CPU usage
requirement, but doesn't your operating system have tools for giving you
that?
--
Philip Hazel University Computing Service,
ph10@??? New Museums Site, Cambridge CB2 3QG,
P.Hazel@??? England. Phone: +44 1223 334714
--
*** Exim information can be found at
http://www.exim.org/ ***
