rrd data capture and multiple tests with missing data
list David Boldt
In a test defined to store 4 values for graphing I am unable to
arrange for the storage of the available information if
information for any variable is missing.
In many runs of the test not all of the variables will have data.
I've tried:
* omitting the variables where there are no values
* listing the variable without data
* setting the value to NaN for missing data
* setting the value to a number below the rrd "--minimum" for the variable
In all of these cases no data is stored for the interval if any of the
variables has no data. (Am using NCV parsing of the status.)
I am able to store and graph the data for all 4 variables if I set missing
values to zero, but this provides unsatisfactory looking graphs and
incorrect averages.
xymon version 4.3.28 on CentOS 7.
-- David Boldt
<user-945c7be1b8e4@xymon.invalid>
"Anybody can sympathise with the sufferings of a friend,
but it requires a very fine nature to sympathise with a friend's
success."
--Oscar Wilde (1854 - 1900)
list Michael T Pins
▸
Boldt, David writes:
In a test defined to store 4 values for graphing I am unable to arrange for the storage of the available information if information for any variable is missing.
In many runs of the test not all of the variables will have data.
I've tried:
* omitting the variables where there are no values * listing the variable without data * setting the value to NaN for missing data * setting the value to a number below the rrd "--minimum" for the variable
In all of these cases no data is stored for the interval if any of the variables has no data. (Am using NCV parsing of the status.)
I am able to store and graph the data for all 4 variables if I set missing values to zero, but this provides unsatisfactory looking graphs and incorrect averages.
xymon version 4.3.28 on CentOS 7.
Use SPLITNCV rather than NCV, then use FNPATTERN in graphs.cfg.