(Aside: this thread has already taken a large chunk of this group's
bandwidth, so this will be my last post on it)
I don't see missing data points - every point has a value. I do see dips
in the graphs. You have chosen to graph:
sum by (foo) (X) / sum by (foo) (Y)
You haven't described what X and Y represent, so I have no idea even if
this is a sensible ratio, but let's assume it is.
If you see dips in the numerator or the denominator, it could be because:
- the individual values are lower
- the number of values being summed is lower - i.e. the number of
timeseries where foo="some value" has gone down.
You can check if it's the latter by graphing:
count by (foo) (X)
count by (foo) (Y)
Let's suppose that the number of timeseries is indeed reducing. I wouldn't
be asking "how can I get rid of the dips in my graphs?" I would be asking:
"What are these dips in the graphs telling me?"
They might be telling you:
1. There's an intermittent problem with the system you're monitoring [if
so, I'd want to find and fix it]
2. There's an intermittent problem with data collection, or the metrics
themselves are bad [ditto]
3. There's an occasional normal event which naturally causes these dips [if
so, I'd want to understand it]
4. The expression I'm graphing is the wrong one, i.e. it's calculating the
wrong value
In any of these cases, I'd want to understand it and if possible fix the
root cause. If you just paper over the cracks in your graphs, you're
pretending the problem doesn't exist.
Regards,
Brian.
--
You received this message because you are subscribed to the Google Groups
"Prometheus Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To view this discussion on the web visit
https://groups.google.com/d/msgid/prometheus-users/09059e52-4d13-47a1-aa17-09cfc675bf61%40googlegroups.com.
