[rrd-users] Re: Monitoring availability with RRD

[Frank Harper]

Steve, thanks for your positive feedback.

I'm glad to see you agree that the amount of "unknown" information, is
important.

The basic assumption that I'm making here, is that when information for a data
source is unavailable that means there is a service outage for the
corresponding network node/application. Of course that might not be true for a
variety of reasons, but I hope the approximation will be useful in managing
our networks.

I'm in the process of writing scripts that will provide this information. They
use rrdtool directly, so, with a little work, they should work with any of the
rrdtool front ends. I'm currently using cricket, so that's what the first
version will work with for sure.

My basic idea is to have a script update an rrd containing availability
information for a set of data sources at regular intervals (1 day default). I
will record two availability numbers: working hours and 24h/7.

A cgi will then use this information to provide the following HTML reports:

Textual statistics, over a certain time period, for all data sources in the
availability rrd, sorted by availability (ascending). This should tell us
where reliability may need improvement.

Graphics for one data source in the availability rrd, over a certain time
period. This will show us what the trend is for a particular data source.

When all this is working, I'll send it to Tobi to be included in the contrib
directory.

steve rader wrote:

>  > From: Frank Harper
>  > I want a textual report containing availability statistics for some my
>  > RRD datasources. Now, it seems to me like this would be a pretty common
>  > thing to do. But strangely enough, I haven't found anything available
>  > that looks like it would fit the bill.
>  >
>  > Did I miss a publicly available tool/contribution that would do this?
>  > Or is this something so trivial that everybody just rolls their own?
>  > Or maybe this isn't even a good idea (please tell me why not)?
>
> Frank,
>
> This is a good idea: I'm thinking I'll write a perl script
> for NRG that would produce a plain text report of (NRG) data
> collection failure over some arbitrary reporting interval
> (default one day.)
>
> So then cron could send this report every night to provide a
> terse report of "monitoring availability" over the last day.
>
> Is that what you're thinkin' of?
>
> If so, maybe send me an example of how you'd like the report
> and I'll cook something up.
>
> If not, tell us more about what you mean by "textual report
> containing availability statistics".
>
> steve
> - - -
> systems & network guy
> high energy physics
> university of wisconsin

--
Frank Harper, TURBOMECA network manager
e-mail: Frank.Harper at turbomeca.fr
Tel: +33-559-12-52-18
Fax: +33-559-12-51-50


--
Unsubscribe mailto:rrd-users-request at list.ee.ethz.ch?subject=unsubscribe
Help        mailto:rrd-users-request at list.ee.ethz.ch?subject=help
Archive     http://www.ee.ethz.ch/~slist/rrd-users