Change bucket color based on threshold

[kan1207]

I am using Kibana to visualize my server status and performance. Though there is watcher, I would like to know if it is possible to have thresholds to implement on the visualized data. For example, the bar chart for Disk Usage is green below 80% usage, Red for above.

[Bargs]

Hi @kan1207, you can already to this with a range filter sub-aggregation. Here's a screenshot of an example:

You can change the colors of the bars by clicking on the labels in the legend.

[kan1207]

Oh sorry, what I mean is to automatically change into another color but not split ranges into 2 bars

[Bargs]

I'm not sure I understand the difference, don't they accomplish the same thing?

[kan1207]

In your figure, that is the count of cpu usage percentage within the range. Instead of spliting the bar into different ranges and count their existance, I hope there is a threshold that can enable to monitor live average cpu usage within a certain time range. so that whenever the average cpu usage reaaches a certain percentage the bar will turn to another color according to the threshold set. For example, the average cpu usage should be around 60% (normal range: green color), when the usage suddenly increases to 90%, the same chart will turn to red color which indicates a warning.

Hope I have made it clear

[Bargs]

when the usage suddenly increases to 90%, the same chart will turn to red color which indicates a warning.

So the entire bar (the one that's above 60%) will turn red, or the entire chart will turn red?

[kan1207]

he entire bar will turn red, sorry for confusing

[Bargs]

Got it. That seems reasonable. There are other ways I think you could achieve the same result, but maybe not quite as nice. You could create a chart that is filtered to only show values above the threshold, so that if you see anything in that chart you know it's bad news. I'll reopen this though since it could be a nice-to-have.

[sharon-ca]

Yes, it is definitely a very good feature for data visualisation.

[tbragin]

@Bargs is this a dup of #4482 ?

[Bargs]

Ah, yep, thanks @tbragin. Closing as dupe.