CPU Utilization is very high

[Rawat_Sunil]

In FAS 3250 Storage , there is very high CPU Utilization please suggest how we can mitigate it.

Please note down that we dont have the Net App Support.

Regards,

[ogra]

Hi Sunil,

It will be great if you have Support. 

High CPU utilisation mostly is due to host load / other process load. I suggest you make sure you have good enough free space in your Aggrs, all the disk spindles are serving IOPs evenly. 

Additionally, make sure you don't want to run processes like Dedupe / SnapMirror all volumes you don't require. 

Any idea what is the total IOPs serviced from the FAS3250 system ? 

Regards,

Bakshana

[rwelshman]

You can gather some statistics of the usage on the filer and try to determine which volume is contributing the most to the usage and investigate from there. As mentioned above sis / dedupe and replication will contribute to the CPU usage as well.

[hariprak]

Hi,

Can you provide us the output of the command "sysstat"

Data ONTAP 8 7-Mode:
sysstat -m displays per-processor and average utilization.
The ANY column in sysstat -m output shows the percentage of the time that one or more CPUs were busy. In addition to this, the utilization of each individual processor is displayed, as well as the average (AVG).

Data ONTAP 8 Cluster-Mode:
statistics show -category processor displays per-processor utilization for each node in the cluster.
statistics show -node <nodename> -category processor displays per-processor utilization for the specified node(s).

Thanks

[RPHELANIN]

Are all CPU cores high or just 1? its not uncommon to see a single CPU run at +98%

[Rawat_Sunil]

How we determine which volume is contributing the most to the usage?

Is there is some command to identify the culprit volume?

Regards,

Sunil Rawat

[Rawat_Sunil]

Thanks to all.

At present the issue is resolved and the main culprit process is the NDMPD.

[heightsnj]

I have same questions below. Can any body please explore more on these two questions . Thanks!

How we determine which volume is contributing the most to the usage?

Is there is some command to identify the culprit volume?

[JGPSHNTAP]

Yes, you can figure out pretty easily which volume is producing the most ops and go from there.

priv set adv

stats show -n 50 -i 1 volume:*:total_ops

Also, high CPU doesn't always mean a bad thing.. check it out with sysstat -M 1