Been plotting things with Grafana with prometheus as datasource. One thing
that i am not really clear about is null and Nan values. I wonder:
- What is the meaning of null in prometheus ?
I can’t answer that because I’m not familiar with Prometheus, but hopefully
someone else can.
- When can a time series return the value null ?
When there is no value (result) for the time period covered by the query.
Either the query has asked for a period when there was not intended to be a
measurement, or maybe a measurement failed or went missing and just isn’t in
- Is the value null the same as NaN ?
I thought NaN (not a number) was something Python-specific (although
https://en.wikipedia.org/wiki/NaN clearly proves me wrong about this), and I’d
say “no, it’s not the same as null” because null means “nothing”, whereas
“NaN” means “not a number”, but it could still be something else which isn’t
“nothing”. “Null” is more specific to me than “NaN”, because “null” means
“there is no information”, whereas “NaN” could mean “oranges”.
- In grafana, graph panel, we can interpret null as zero, as null, or
connected ? What does that mean exactly ?
“Null as zero” means “if you get a null value, substitute zero and use that as
the value for this time sample”.
“Null as null” means “if you get a null value, substitute nothing - it’s
null”. Most graphs will leave a gap between the preceding and succeeding data
points in this case.
“Connected” means “if you get a null value, join the preceding and succeeding
values with a straight line, effectively pretenting that you didn’t expect to
get a measurement where the null appeared”.