Thold Behavior Question

[yousillygoose]

Hey All,

I have a question for how Thold will handle something. Assume you are monitoring a volume 'D' and want to alert if it breaches 90% capacity (Hi/lo threshold).

All is well and good and the device doesn't breach the threshold. Let's assume D is unmounted (for whatever reason). To my knowledge, the datasource will fill with NaNs. When this happens, the threshold may or may not have been breached. Does thold assume all is healthy or will it alert in this situation (it seems that from experience it doesn't alert).

The problem we have is that datasources break occasionally that we assume we are alerting on. We have a very large cacti environment so it isn't possible to manually monitor the datasources.

So, long story short: if a datasource is storing NaN's and thold is doing a Hi/Low threshold- will it alert?

If not, is there a solution? I imagine a script would need to be written that goes through all datasources that have an associated threshold that is enabled to ensure that in the past 'x' time period the datasource had a value.

Thanks for the help.

[TheWitness]

I would first look at the HMIB plugin, and then write some extension to perform some key checks. The problem is that volumes come and go, especially NFS. So its a bit tricky to do things en mass.

Second option is to have thold option to alarm on nan.

[yousillygoose]

Hmm- I was unaware that you can make 2 different thresholds for the same datasource. How do you alert if the value is NaN? Is there an exactly value you can compare against (I imagine using "NaN" probably doesn't work) or is there an RPN perhaps?

[TheWitness]

Boy, Androids auto correct can slaughter a typed message. You might want to read my message again. Have to stop using Android....

[Hanseat]

How I got it to work for me: http://forums.cacti.net/viewtopic.php?p=251781#p251781