[Nagiosplug-devel] Best practice for thresholding

[John Vincent]

On Mon, Jun 22, 2009 at 4:03 PM, John Vincent<nagiosplug-devel at lusis.org> wrote:
> On Mon, Jun 22, 2009 at 3:42 PM, Ton Voon<ton.voon at opsera.com> wrote:
>>
>> On 22 Jun 2009, at 18:22, John Vincent wrote:
>>
>>> So I've been attempting to use check_memcached and in the process was
>>> cleaning up some invalid perfdata output. I also noticed that the
>>> thresholding is somewhat broken. This plugin, like many database
>>> plugins, has a hit ratio metric. Hit ratio metrics make the most sense
>>> when approcahed from a "warning/critical below X%" (at least in my
>>> mind). So if my hit ratio drops below 75%, I want a warning and if it
>>> goes below 50% I want a critical.
>>>
>>> What's the best way to handle this with Nagios::Plugin? Should the
>>> logic be done in the script or should it be done in the module itself?
>>
>> Nagios::Plugin already supports the thresholds defined in the guidelines:
>> http://nagiosplug.sourceforge.net/developer-guidelines.html#THRESHOLDFORMAT
>>
>> So you should already be able to use the threshold routines in N::P.
>>
>> If it doesn't work properly, then that would be a bug.
>>
>> Ton
>>
>>
>
> Ahh okay. Thanks Ton. I wasn't looking hard enough to find what I
> needed. I was simply looking at it from a greater than/less than
> scenario as opposed to being outside a given range.
>

One question that I have regarding these threshold ranges. Should the
range definitions be passed as-is to the perfdata output or not? If
so, then pnp is at fault for not handling it properly in this case (or
I really need to make a custom template to graph it properly.).
Currently pnp is breaking on "hits=90%;@30:50;30" due to the format of
the warn value. Just trying to figure out where to go with it.