Monitor ops with harvest | NetApp | Page 1

sharp zodiac Mar 5, 2024, 5:33 AM

#

HI,
I am trying to monitor volume ops with harvest at volume layer.
Here are metrics that we use:
use "qos_ops" if the volume has qos policy.
use "volume_total_ops" if the volume doesn't apply qos policy
Is this correct?

novel wharf Mar 5, 2024, 10:54 AM

#

@sharp zodiac Burt 440535 contains some information on this topic, but it's quite old. @wet marsh might be able to provide further insights.

sharp zodiac Mar 7, 2024, 1:12 AM

#

@wet marsh Is the metric that I used correct?
Since that the value I got from this metric "qos_ops" has a gap to the value show in AIQ Unified Manage, and also with some exceptional spike value.
And I used "volume_total_ops" instead, it seems normal.

novel wharf Mar 7, 2024, 6:02 AM

#

@sharp zodiac Probably asking in the #1062049169520476220 channel is also a good idea. Also, spike values are likely a bug that we are triaging in another thread.

sharp zodiac Mar 8, 2024, 4:55 AM

#

Thanks!!

wet marsh Mar 8, 2024, 3:24 PM

#

Yes?

#

I'm not 100% sure which counter Harvest is using but there are two objects from ONTAP used: workload objects (workload/workload_queue_nblade/etc.) and volume.

#

Volume only measures WAFL latency but workload would track QoS latency.

#

Is that the queestion?

#

If you want please send an e-mail, first dot last at my company

#Monitor ops with harvest