fervent knot
Hi,
we are monitoring cpu utilization by using "node_cpu_busy"
And from image1, you can see the cpu utilization in harvest is keep growing.
And we collect the performance archive to find the reason for this increase. But the value in PA has no significant increase. And the description of that PA is in image2. I suppose they are using the same metric?
@fathom wren Please help to confirm this issue? Thanks
fathom wren
@fervent knot
Could you compare metric value node_cpu_busy with below ONTAP CLI command in diag mode?
They do match for me.
statistics show-periodic -interval 60 -iterations 10 -object system:node -counter cpu_busy -instance NODE_NAME
fervent knot
@fathom wren Do you know if the PA using the same metric?
fathom wren
I am not sure about PA value. CLI mapping I gave above is the value represented by Harvest.
fervent knot
@fathom wren Is there any possible reason that only this cluster has increasing value at that period? We checked the AIQUM and PA, but the value has no significant increase.
We also checked the CLI, and it does map to the harvest value after the value is lower down.
fathom wren
@fervent knot We are not aware of any bugs in this area. What is the Harvest version?
fervent knot
harvest 24.05
But only one cluster shows this increment,
others are normal.
So not knowing how to explain the increment during that period
fathom wren
Let's monitor this cluster via the CLI I shared above for cpu_busy data for a couple of days and observe the values against Harvest. If the values deviate significantly, we will know if there is a bug in Harvest.
solid rock
What is PA ASUP ID? I'd be glad to look.
fervent knot
What is PA ASUP ID? I'd be glad to look.
ASUP ID: 2024091023480220
Here is the link to PAS:
https://pas.activeiq.netapp.com/grafana/d/cpu-view/cpu-view?orgId=3&from=1725753600000&to=1726271999000&var-customer_name=Taiwan Semiconductor Manufacturing Co Lt d.&var-cluster_name_uuid=f18p4nasnac1 (9ccfce71-2e4c-11eb-8576-d039ea235ecd)&var-node=f18p4nasnac1n01
From the PAS, we don't see a significant increase in CPU utilization.
But from harvest, we do see that increase.
fervent knot
@solid rock Yes, but from the PA CPU utilization table, the average line doesn't have significant increase like what we see in harvest
solid rock
Yeah. @fathom wren or @kind spindle which counter is being used for Harvest?
fathom wren
Harvest uses similiar metric as presented by below ONTAP CLI
statistics show-periodic -interval 60 -iterations 10 -object system:node -counter cpu_busy -instance NODE_NAME
