[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [afnog] [Mailer-Daemon at mantse.gh.com: Mail delivery failed:returning message to sender]
- To: Stephane Bortzmeyer <bortzmeyer at nic.fr>
- Subject: Re: [afnog] [Mailer-Daemon at mantse.gh.com: Mail delivery failed:returning message to sender]
- From: Brian Candler <B.Candler at pobox.com>
- Date: Mon, 13 Oct 2003 09:50:56 +0100
- Cc: afnog at afnog.org
- Content-Disposition: inline
- Content-Type: text/plain; charset=us-ascii
- Delivered-To: afnog-archive at lists.eahd.or.ug
- Delivered-To: afnog at afnog.org
- In-Reply-To: <20031013084104.GA24530 at nic.fr>
- List-Archive: <http://listserv4.cfi.co.ug/pipermail/afnog>
- List-Help: <mailto:afnog-request at afnog.org?subject=help>
- List-Id: The AfNOG general discussion list <afnog.afnog.org>
- List-Post: <mailto:afnog at afnog.org>
- List-Subscribe: <http://listserv4.cfi.co.ug/mailman/listinfo/afnog>,<mailto:afnog-request at afnog.org?subject=subscribe>
- List-Unsubscribe: <http://listserv4.cfi.co.ug/mailman/listinfo/afnog>,<mailto:afnog-request at afnog.org?subject=unsubscribe>
- References: <20031010203232.GA38454 at uk.tiscali.com><20031013084104.GA24530 at nic.fr>
- Sender: afnog-bounces at afnog.org
- User-Agent: Mutt/1.4.1i
On Mon, Oct 13, 2003 at 10:41:04AM +0200, Stephane Bortzmeyer wrote:
> On Fri, Oct 10, 2003 at 09:32:32PM +0100,
> Brian Candler <B.Candler at pobox.com> wrote
> a message of 126 lines which said:
>
> > Run it from cron every 10 minutes.
>
> VERY bad idea, IMHO. It will fill up your disk even faster. The
> problem is that cron has no memory of the previous run. If cron
> launches the script every ten minutes and the problem lasts two hours
> during the lunch, you get twelve messages for nothing.
12 x 2KB = 24KB. And if the mails are successfully delivered there will be
no disk space used apart from a few maillog entries.
> Instead, you should use a proper scheduler like mon
> <URL:http://wwwkernel.org/sofwtare/mon/> which can:
>
> * trigger the alarm only after N failures,
> * only send the alarm every M minutes even when you monitor every N
> minutes (with M >> N),
> * stop sending the alarms after N attempts (sort of automatically shut
> down the alarm).
Certainly, you can install something fancy and complicated if you like.
Equally, you could argue that you should have snmpd running and a
centralised NOC workstation which logs and graphs every variable on your
system. But I'm quite happy to install a single file which needs no extra
software support (and I'd install it in addition to the snmp monitoring, if
you have it)
A basic script which just sends an E-mail when your disk utilisation hits
90% on any partition is a real life-saver. The point of this script is to
start sending mails *before* the system gets to a critical condition.
Regards,
Brian.
__________________________________________________
This is the Africa Network Operators' Group(AfNOG)
technical discussion list.
The AfNOG website is: <http://www.afnog.org>