#Calculate the total ops for a cluster

1 messages · Page 1 of 1 (latest)

agile wharf
#

Hi,
I am trying to get the total ops for a cluster
and here are two metrics:

sum(volume_total_ops{datacenter=~"$Datacenter",cluster=~"$Cluster"}) by (cluster)

sum(qos_ops{datacenter=~"$Datacenter",cluster=~"$Cluster"}) by (cluster)

I assume that these two value would be similar?
And volume_total_ops might be a little higher
than qos_ops since it include some system ops.
Is this explanation correct?

But from user site,
sum(volume_total_ops{datacenter=~"$Datacenter",cluster=~"$Cluster"}) by (cluster) has the value of 19K
sum(qos_ops{datacenter=~"$Datacenter",cluster=~"$Cluster"}) by (cluster) has the value of 29K
There is a bug gap between this two value
Is it because qos_ops should exclude all workload with {policy_group="User-Best-Effort"}?

tribal pilot
#

@agile wharf I think qos_ops will be higher. See if explaination here helps
https://community.netapp.com/t5/Active-IQ-Unified-Manager-Discussions/difference-of-quot-qos-ops-quot-and-quot-total-ops-quot/m-p/118132/highlight/true#M21008
@acoustic marten can help as well.

Re: difference of "qos_ops" and "total_ops"

Hi ,   Good question!  total_ops tracks WAFL operations while qos_ops tracks protocol operations.  WAFL IOs have a max size of 64KB but client IOs can be much larger.  So if a 128KB protocol IO arrives it will cause 2 x 64KB WAFL IOs and you will see the counters deviate.  Both qos and normal volum...

acoustic marten
#

Hmm. I wonder which statistics counter/object is compared?

#

Oh this is another thread. Yeah this is expected.

#

The volume object is only WAFL/dblade ops. The workload* object has more visibility into the nblade across the cluster. There may be repeat operations being rewound, but I'm not 100% sure how it handles that. IOP size might be different too, but that may be different too.