NetApp Lun Latency

[itsmemattchung]

Hey everyone,
This is my first post and I look forward to learning and contributing to forum. Recently I have been assigned the task of creating a dashboard for my NOC. After taking some time to learn how to use rrdtool and cacti, I have been steadily making progress.

Currently, I am working on establishing baseline statistics for our NetApp filers (3170, 3070, and 3020). I am now on grabbing lun statistics (primarily latency/IO) using NetApp's 4.0 SDK. I did find a user on the forum who supplied his/her script, however the latency metrics are miscalculated. Here is the issue I am currently running into:

LUN latency formula for the NetApp is (avg_latency[2] - avg_latency[1])/(total_ops[2] - total_ops[1])

In order for the script to appropriately work, I am grabing the first set of counters, pausing for 1 second, grabbing the second set of counters and performing the calculations. The script has been working great, but I've noticed that after too many data sources on cacti, data doesn't not get collected for the majority of the luns (There are approximately 80 luns). I then removed all the data sources and have been slowing adding one data source at a time.

I have been reviewing the logs and have been reading through the forums (upgrading to newest version, using spine, etc) but have not found anything that deals with scripts pausing for certain intervals. I'm hoping that once this gets resolved, I can post the data query/graph templates along with the script so everyone can baseline as well.

If someone could point me in the right direction, I would really appreciate it. Thank you!

[itsmemattchung]

Actually now that I think about it, this should be possible by using the DERIVE counter for both dataset, and using a CDEF. I'm going to give it a shot and hopefully be able to share the templates.

[Sirup]

Maybe this helps you: http://forums.cacti.net/about8748.html