Some graphs are failing other not, same config

[WillemDH]

Hello,

I recently changed the way I monitor CPU usage for Windows servers from check_nt to nrpe. So now I noticed some services have nice graphs and others don't., while they are all based on the same policy and created around the same time.
What could be the reason some of them are working fine and others aren't? The difference seems to the bad ones only gather the 1-minute average data and the others also gather the 5-minute and 15-minute data.
I include two screenshots to show the difference:

In the last screenshot you can see he doesn't seem to find the perfdata for 5 and 15 minutes.

[abrist]

Changing metrics, or even worse, adding or removing metrics from performance data can lead to all sorts of oddities in your graphs. You will most likely have to remove the rrds and xml files associated with these checks. They will be regenerated correctly then after a few minutes. You will lose the previous historical information though.

You will find the rrds in the directory:

Code: Select all

/usr/local/nagios/share/perfdata/<hostname>

Remove the broken rrds/xmls for each of the broken services for each host in question.

[WillemDH]

Removed the rrd's and it seemed to work after that. Strange because some worked immediately after cloning the service while others with +- exact the same config didn't work. Anyway it works now.

[sreinhardt]

Glad to hear its working, to add to abrist's note even if they are using the same amount of tracks or less than the previous rrd, if the data is returned in different increments or percentages instead of mhz, that would likely be the cause of the changes. As you did, recreating the rrd is about all that can be done without some serious conversion.

[WillemDH]

Spenser, Andy,

This thread can be closed.

Grtz