2009-02-10 09:29 AM
I have had a customer ask the following question:
Could you find out a good description of what nfs_latency is in Ops Mgr?
I had this from a previous request yet not sure it is the same:
You can expect PA to grab roughly 5% maybe more if you turn on a lot of additional counters. If you don’t need to look at all of the filers for this issue you can use the command: dfm host set <filername> perfAdvisorTransport=disabled to not collect data on one or more. (The performance data and charts we have show ~2% overhead with a few peaks up to 4.5% CPU utilization)
If you want to exclude a lot of Filers put them in a group and use dfm host set <groupname> perfAdvisorTransport
Can you let me know a “good description of what nfs_latency is in Ops Mgr?”
I look forward to your response.
2009-02-11 02:56 PM
Here is a screen shot of the filer:
The nfs_latency show huge spikes and the customer does not know why. This doesn’t look like an average of the other nfs categories…
The latency looks bad and not sure why or what to look at…
2009-02-12 10:31 AM
I looked up the BURT and see that it states that this is fixed yet I don't see that 3.8 is released? Has the new version been released? My customer has an N-Series and can't find DFM 3.8 on the IBM web site...
Can you let me know if the new release is available and where I can point my customer to?
2009-02-12 01:40 PM
As you have found out, DFM 3.8 server version has not been released by NetApp yet. Typically IBM delivers the software for N series in reasonable currency (within 2-4 weeks of when we make it available to them), so correspondingly, DFM v 3.8 is not available for N series either. That said, I would encourage you to point the N series customer to have that conversation with the IBM support team if this is causing business issues as they would be the right folks to assess the total problem and appropriate resolution.