stack limit in histogram?

This support forum board is for support questions relating to Nagios Log Server, our solution for managing and monitoring critical log data.
User avatar
hsmith
Agent Smith
Posts: 3539
Joined: Thu Jul 30, 2015 11:09 am
Location: 127.0.0.1
Contact:

Re: stack limit in histogram?

Post by hsmith »

Is that going to be your talk at NWC?
Former Nagios Employee.
me.
_asp_
Posts: 91
Joined: Mon May 23, 2016 4:30 am

Re: stack limit in histogram?

Post by _asp_ »

strange. If I enable fill zero the stack problem is going away.

But what I do not understand:
I suppose only to be items considered to mean calculation where the field is present. Even if I set the interval to 5 minutes where **always** multiple matches are present, the system should be able to calculate the mean value.

So I have a valid value all the time, kibana should be able plot all lines without having to deal with null values.
User avatar
hsmith
Agent Smith
Posts: 3539
Joined: Thu Jul 30, 2015 11:09 am
Location: 127.0.0.1
Contact:

Re: stack limit in histogram?

Post by hsmith »

From what it sounds like, it's working as intended. You can't calculate a mean on a null. I don't know if this is something we're going to be able to change/fix.
Former Nagios Employee.
me.
_asp_
Posts: 91
Joined: Mon May 23, 2016 4:30 am

Re: stack limit in histogram?

Post by _asp_ »

but I do not understand, where the value null is coming from.

see, lets say, i have following events with following values:

minute: value
1: 10
3: 0
4: 6
5: 3
6: 0
8: 0
9: 21
10: 45
11: 54
13: 0
14: 15

That's how it looks like in detail:
We have a probe each minute, each 5 minute the whole event is missing, not only the field.
So you have missing events on minutes 2, 7, 12.

If I go for 1 minute interval, I agree to you, that the algorythm will have intervals, which have no values. That might cause plotting problems.
But If i set interval to 5 minutes, than it allways have valid values:
interval 1 (minutes 0-5 = (10+0+6+3)/4)) -> mean of 4 events
interval 2 (minutes 6-10 = (0+0+21+45)/4) -> mean of four events

That's how I expect it to work and see no problem here as long as there are events in the interval size set in panel.
Also I'd expect - if fill zero is not checked - that the line has gaps for the intervals where the mean value could not be processed (e.g. by division by zero) or that the gaps are replaced by a direct line between previous and next valid values.
User avatar
hsmith
Agent Smith
Posts: 3539
Joined: Thu Jul 30, 2015 11:09 am
Location: 127.0.0.1
Contact:

Re: stack limit in histogram?

Post by hsmith »

I can put in a feature request to change how it functions when zero fill is not checked, but I don't know if this is something that would be implemented or not.
Former Nagios Employee.
me.
_asp_
Posts: 91
Joined: Mon May 23, 2016 4:30 am

Re: stack limit in histogram?

Post by _asp_ »

In my opinion its a bug and not a feature request.
But if you do not accept it as bug, please add a feature request.
User avatar
hsmith
Agent Smith
Posts: 3539
Joined: Thu Jul 30, 2015 11:09 am
Location: 127.0.0.1
Contact:

Re: stack limit in histogram?

Post by hsmith »

swilkerson is out of the office for a couple more days. When he returns I'll see what he deems appropriate for this once, as he was working with you on it a bit.
Former Nagios Employee.
me.
Locked