Forum Discussion

Eric_Singer's avatar
8 years ago

No data threshold

Hi,

I love the idea behind the "no data" warning, but it's missing one thing that I think would help make it a little more friendly.  We have a few devices that have spotty collection reliability, and it often times leads to use seeing "no data" warnings blowing up in our alerts console for a few minutes and then disappearing.  I'd really love to see an option similar to what exists with the actual data points, that is, only show a "no data" if there's "x" consecutive polls of no data.  This would prevent the times where we might miss two polls in a row from flooding the alerts.  

  • I have another post elsewhere additionally requesting this.   GLAD to see great minds think alike
    .

     

     

  • The alert trigger intervals work for no data alerts, too. For a few reasons, it's not good practice to have a no data alert set on a datapoint that also has a value-based threshold. (For one, the message should definitely be different) Is that what you're talking about?

  • On 12/13/2016 at 1:44 PM, Mike Aracic said:

    The alert trigger intervals work for no data alerts, too. For a few reasons, it's not good practice to have a no data alert set on a datapoint that also has a value-based threshold. (For one, the message should definitely be different) Is that what you're talking about?

     

    Mike, are you saying that if we set Alert trigger interval (consecutive polls), then the no data alert will honor that as well?
    IE if trigger interval is set to 2, the no data alarm would only happen if the collector receives no data after three consecutive polls??

  • Mike,

    Can you expand on this a little further?  For example, if i have something simple, like a CPU alert, and its set to trigger after 3 consecutive polls.  Are you saying if there's no data for 2 intervals, it wouldn't display a warning, but after the 3rd interval in a row it will?

    Just to restate the situation I'm in.  I have a few devices that just don't poll reliably, they probably have a "no data" every couple of polls.  I would love the ability for those types of devices to set something like "if there are 3 polls in a row of no data, throw a no data warning" and clear instantly.

  • Hello Everyone,

    I just saw this and wanted  to confirm what Mike was referring to. If I have a datapoint, such as CPUBusyPercent, that has an "Alert trigger interval (consecutive polls)" set to 4 AND I set "If there is no data for this datapoint:" to trigger an alert, the No Data alert would trigger after 4 consecutive polls, just as the regular thresholds would that are set on the datapoint. 

    If you wanted to extend the time to trigger a No Data alert, or have a specific amount of consecutive polls for No Data, you can always set the No Data alert on a datapoint that has no threshold, adjust the "Alert trigger interval (consecutive polls)" to what you deem appropriate, and it will respect the amount of consecutive polls set before triggering a No Data alert. 

    For the most part this can be any datapoint you choose on the datasource, as usually when you are getting No Data, it is not just that one specific datapoint, it is usually the whole protocol not responding (WMI, SNMP, JDBC, etc.).

     

    Hope that clears this up, let us know if you have any other questions related to this.