Problem with 95th percentile code

[rkuris]

Hi,

I found a problem where if there is ever a "nan" in the output from rrd for your data source on the interval in which you are scanning data, even if it is on a data source you are not interested in, the data will not be recorded in the 95th percentile scan functions.

In my particular case, I had some other data in the rrd file that always had a "nan" in one of the columns. This prevented ANY 95th percentile calculations on the other columns.

I made a quick fix for this by adding "$output = str_replace('nan','0.0e+00',$output);" just after the $line_one assignment in rrd.php. However, that isn't the right fix; the correct fix is to make the regexps further down match "nan" as well as a data value.

Since I know my patch can possibly produce wrong data, I don't want to include it as an easy to apply patch, but I figured someone here might want to take a look at the regex code and come up with some way to match 'nan' as well.

[rony]

Nth percentile code is a beast....

That being said, can you please submit a bug on this matter. I will take a look at it, but will not promise any changes.

http://www.cacti.net/bugs.php

Don't forget to reference this post in the bug. But don't make that your bug report. Details should be in the bug report.