No performance data saved if number of variables go down

[DennisPR]

We have created a script that reads the IOPS per volume for our NetApp controllers.
The script returns the iops per volume as data and as performance data and has been working perfect for a couple of months eg :

Code: Select all

vol0=7 nfs_iit=361 nfs_dev02=162 nfs_temp=0 | vol0=7 nfs_iit=361 nfs_dev02=162  nfs_temp=0

Today we have seen that some of the services haven't been adding any data to the performance grahps since a couple of weeks.
The problem seems to occur on the controllers were we have deleted some volumes eg :

Code: Select all

vol0=7 nfs_iit=361 nfs_dev02=162 nfs_temp=0 | vol0=7 nfs_iit=361 nfs_dev02=162  nfs_temp=0
vol0=7 nfs_iit=564 nfs_dev02=212 nfs_temp=0 | vol0=7 nfs_iit=564 nfs_dev02=212 nfs_temp=0
vol0=7 nfs_iit=132 nfs_temp=0 | vol0=7 nfs_iit=132 nfs_ nfs_temp=0

If we delete one of the volumes, the script returns data for the remaining volumes but the performance graphs fail.
It shows vol0, nfs_iit, nfs_temp & series 4 and does not save the performance data anymore until we add a 4th volume again.

Here's an example graph

Netapp.PNG

I'm pretty sure that the problem is not related to our script.
Has anyone seen this with any other service that has a variable number of results returned as performance data ?

[abrist]

This is due to the nature of rrds. They are static bins with a fixed number of datasources. We highly suggest that you configure each mount point check as a separate check to avoid this issue. I wrote a script to add missing datasources to fix some issues with the 2014 upgrade. I will look into removing datasources as well.

[DennisPR]

Thx ! We've added individual tests per volume instead of 1 test for all.