Too many timeseries, filter on values

[ReneVisser]

We have create a lot of custom metrics based on sensu monitoring. These metrics can have a value of either 0 (ok), 1 (warning), or 2 (critical)
I want to build a chart base on a query like 
data('sensu_*', filter=filter('sf_environment', 'prd') and (filter('owner', 'teamA') or filter('type','platform'))

this query returns over 3000 timeseries so results are sampled:
(Rendering 1000 out of 3270 time series. Values displayed may not be representative; consider filtering your chart or applying aggregation analytics.)

In the end I want to have something like this:
A = data('sensu_*', filter=filter('sf_environment', 'prd') and (filter('owner', 'teamA') or filter('type','platform')), rollup='max').between(1,3).count(by='host.name').above(0).publish(label='Critical', enable=True)

but now the result from this query isn't trustworthy.


There no extra dimensions to filter upon, but I want to filter all the results where value is 0.
when I use methods like above(), between or when i create a new stream with when() I keep getting this warning.
Does anyone have a suggestion what I can try.

[bishida]

Hi,
The wildcard * in the metric name may be causing multiple MTS to be returned and all being represented as "sensu_*". Maybe try specifying a single metric name and see if that helps. If you have custom metrics in addition to other sensu metrics, you may need to create a way to filter just on those custom metrics (e.g., alter the name like sensu_mycustom_* or add a dimension that indicates it's custom)

[ReneVisser]

Thank you for the suggestion,
But we need all the sensu checks to be monitord, so we need all the metrics.
I did however find a dimension which helped me out, coincidentally there was an output dimension which contains "Ok" for all the datapoints containing a 0 as a value for all the metrics. So I was able to. use that.

But still I am wondering if it somehow is possible to filter an the value of metrics.