Here I graphed some data with a gap. I simply generated a random value every 10 seconds and stored it in InfluxDB, turned it off for a while and then back on.
As you can see the missing data is linearly interpolated. That is misleading. The value wasn’t 65 at 04:19:00, instead we simply don’t know the value.
I then found the “Disconnect values” setting and I set it to “> 11s” because I knew I normally had data points at approximately every 10 s. At first glance it worked pretty well, the gap was now rendered as a gap:
Unfortunately this breaks down when I zoom out far enough that the interval gets adjusted. When the interval goes to 15s, suddenly all of the points are disconnected of course because they are now all more than 11 s apart:
(Here I simply adjusted the interval instead of zooming out, but the effect is the same.)
At first I thought since I’m working with the InfluxDB data source anyway (this might not be possible with other data sources) I can use the
createEmpty parameter to fill the gaps with null. This actually works for the longer intervals but this time it breaks down for the short intervals because I then have 9 seconds of null and one second with a value, disconnecting the points again.
How to handle this properly?
Can I maybe somehow use
max(v.windowPeriod, 11s) as the “disconnect values” threshold?